Karyotype and gene panels

In brief

Two tests, two different questions

A karyotype is a photograph. A sample of living cells is persuaded to divide, stopped in mid-division, dropped onto a slide so the chromosomes spread apart, stained so that each one carries a reproducible pattern of light and dark bands, and then photographed, counted and arranged by size ^1,2,7. The result is a picture of the whole genome at very low magnification: 46 objects, and whether any of them is missing, doubled, or has swapped an arm with another. It was the first test in medicine to look at all of a person's DNA at once, and it does so by looking at its packaging rather than its sequence; microarrays and genome sequencing now survey the whole genome too ^37,39.

A gene panel is a list. DNA is extracted from the specimen — it does not need to come from living cells — and a predetermined set of human genes is copied out and sequenced, then compared letter by letter with the human reference sequence. The lists are not small, but they are lists: hotspot regions of 50 genes in a widely used assay, 287 genes in another, 341 in a third, 523 in a fourth ^3,10,11,12. A hotspot is one of the handful of exact positions where cancer-causing mutations cluster: the Ras genes, for instance, are activated by mutation at the twelfth, thirteenth or sixty-first codon — codons being the three-letter DNA words that each specify one building block of a protein ⁴¹. A hotspot assay sequences those frequently mutated regions and not the rest of each gene ¹⁰.

The two tests divide the work of oncology between them along a line drawn by the specimen. Leukemia and lymphoma arrive as fresh blood, marrow or lymph node, full of dividing cells, and the American College of Medical Genetics and Genomics still publishes technical standards for their cytogenetic analysis, the study of their chromosomes ³¹. Solid tumors arrive fixed in formaldehyde and set in wax, and the European and American recommendations that govern them are written entirely in the language of genes and panels ^4,42. One clinical validation shows the two specimen types side by side in a single laboratory: of 214 consecutive cases run on an RNA fusion panel, 73 were leukemia or lymphoma specimens and 141 were formalin-fixed, paraffin-embedded solid tumors, and conventional karyotyping was among the comparator tests run in parallel ⁹.

• Conventional karyotype (G-banding)Cultured cells are arrested in metaphase, banded with trypsin and Giemsa and read down a microscope ^2,7.Still written into leukemia and lymphoma practice, where fresh blood, marrow and lymph node are available ^31,32Sees whole-chromosome gains and losses, translocations (pieces swapped between chromosomes) and large rearrangements at 250 to 850 bands per haploid set ⁸; optical genome mapping, which images very long single DNA molecules tagged at known sequences to map their structure, detected every clinically reported aberration in 36 simple cases, a sensitivity of 100% in the technical comparison, meaning it caught every one ⁵³; in 100 acute myeloid leukemias in 2023 it recovered every reported clinically relevant variant above 5% clone fraction and added findings in 13% ⁵⁴.Submicroscopic imbalances — yield about 3% against 15% to 20% for microarray in developmental disorders ³⁷ — and anything in a specimen with no dividing cells, which a karyotype cannot be made from.
• FISH (fluorescence in situ hybridization)A labeled DNA probe is hybridized to the specimen and counted by fluorescence ⁴³.Where one named rearrangement or gene copy number decides treatment, including HER2, a growth-signal gene amplified in some breast cancers ⁶⁸High for the target it was made for; it works on interphase nuclei, so the cells need not be dividing ⁴³.Everything outside the probe. In the American Society of Clinical Oncology and College of American Pathologists breast guideline, about 5% of results with the two-color probe fall into groups that cannot be called without also reading immunohistochemistry, the antibody stain that shows the protein in the tissue ⁶⁸.
• Chromosomal microarrayPatient DNA is compared with reference DNA across the genome to map copy number, the principle first shown by comparative genomic hybridization in 1992 ⁵⁰.First-tier cytogenetic test for developmental disability and congenital anomalies, and a copy-number tool in tumors ³⁷Diagnostic yield 15% to 20% versus roughly 3% for a G-banded karyotype, from 33 studies and 21,698 patients ³⁷.Rearrangements that move material without changing its amount, and low-level mosaicism, where only a small share of cells carry the change, are generally not detectable, together under 1% of causes in that population ³⁷.
• Targeted next-generation sequencing panelA fixed gene list is captured from extracted DNA and read by massively parallel ('next-generation') sequencing, which reads millions of fragments at once; sizes run from hotspot regions of 50 genes to 523 genes ^10,12.The governing European recommendation names tumor sequencing in advanced non-squamous non-small-cell lung, prostate, colorectal and ovarian cancer and cholangiocarcinoma, and now breast cancer and several rare tumors ⁴95% to 99% sensitivity across alteration types with a positive predictive value — the share of positive calls that are true — above 99% in one validation of 287 genes; detection limits near 2% of reads for hotspots ^3,11.Genes not on the list, and sequence not in the human reference. No published validation has run such a panel against helminth (worm) or protozoan (single-celled parasite) DNA and reported the result.
• Liquid biopsy (circulating tumor DNA)Cell-free DNA — fragments floating in plasma, the liquid part of blood — is sequenced in place of tissue ⁶⁹.May be used when tissue testing is unsafe, unfeasible or too slow, or when a drug's approval allows it; a negative, inconclusive or clinically inconsistent result should be confirmed in tissue, and ctDNA fraction is not a surrogate for disease burden ^69,70Good for point mutations (single-letter changes) in adequately sensitive assays; deep sequencing of 508 genes found that 53.2% of the mutations in patients' plasma, and 81.6% in cancer-free controls, had features consistent with clonal hematopoiesis ⁷¹.Fusions (two genes joined by a rearrangement) and copy-number change are poorly detected ⁶⁹. Parasite DNA does circulate — cell-free Schistosoma DNA was found in the blood of 6.98% of 186 Egyptian patients with endocarditis, infection of the heart's lining and valves — but only by a species-specific assay built to look for it ¹²¹.

From counting chromosomes to naming genes

Human cytogenetics began with a correction. For more than thirty years the human chromosome number was believed to be 48, and in 1955 and 1956 work in Lund established that it is 46 — a finding that required several technical advances at once and, as the historians of the episode point out, required overcoming the power of an accepted conclusion to shape the reading of new results ^5,6. Banding arrived fifteen years later: fluorescent quinacrine in 1970, then a rapid trypsin method in 1971 that could be done in any laboratory, then higher-resolution preparations through the 1970s ^1,2,46. By the late 1980s the same chromosome could be prepared at a 400-band, a 550-band or an 800- to 850-band stage, and routine work spans roughly 250 to 850 bands per haploid set, one copy of each of the 23 chromosomes, ^7,8.

Cancer cytogenetics then produced one of the most consequential pictures in oncology. A small abnormal chromosome was described in chronic myelogenous leukemia in 1960 ⁴⁷, and in the early 1970s, using the new banding stains, Janet Rowley showed that it was not a deletion but a reciprocal exchange between chromosomes 9 and 22 ^48,49. That exchange was later resolved to a fusion gene — two genes joined into one by the swap — and the fusion gene to a drug ⁴⁹. The lesson of that sequence is the one this page keeps returning to: the chromosome picture located the problem, and then a targeted test replaced the picture.

The replacement has been running for three decades and is not finished. Probes made chromosome questions answerable in non-dividing cells from 1986 ⁴³; copy number became measurable across the whole genome from 1992 ⁵⁰; capture of the exome, the 1% to 2% of the genome that codes for protein, made it practical in 2009 to read all of it ⁵¹; the first whole cancer genome had been read the year before, from an acute myeloid leukemia whose karyotype was normal, and it found ten genes with acquired mutations, eight of them never before implicated — the chromosome picture had nothing to show and the sequence did ⁵²; and clinical panels followed ^3,11. By 2021 a single optical mapping assay, which images very long DNA molecules tagged at known sequences, detected every clinically reported aberration in 36 simple hematological-malignancy cases, was largely concordant in 16 complex ones, and showed 100% sensitivity and a positive predictive value (the share of its calls that were true) above 80% against standard tests; the authors begin from the fact that comprehensive structural analysis currently needs karyotyping, FISH and microarrays together, and conclude that optical mapping has the potential to replace them ⁵³. By 2023 a multicenter study of 100 acute myeloid leukemia cases found that optical mapping recovered every clinically relevant variant the standard methods had reported when the clone was present above 5%, found clinically relevant information in 13% of cases that routine methods had missed, and would have altered management in 4% ⁵⁴; a consortium of early adopters published an implementation framework in 2024, with one author from the instrument's maker ⁵⁵, and in 2025 expert recommendations for integrating the method as a standard-of-care cytogenetic assay ⁵⁶. The honest summary is that chromosomes still matter in blood cancers and carry a shrinking share of the decision.

What each method physically does, and what it cannot do

Every one of these tests is a physical process with a built-in blind spot, and the blind spots are not secrets. Banding depends on stain chemistry along a condensed chromosome, so its unit of resolution is a band, not a base ^1,2. FISH depends on a labeled piece of DNA finding its complement, so it answers about the probe's target and nothing else ⁴³. A microarray compares the amount of DNA at thousands of positions with a reference, so it finds gains and losses and is blind to a rearrangement that moves material without changing its quantity: balanced rearrangements and low-level mosaicism are generally not detectable by arrays ³⁷. Sequencing reads short fragments and places them on the reference, so what it reports depends on what the reference contains.

Panels add a second filter on top of the first. Hybridization capture pulls down the regions on the list; amplification-based assays copy them with primers. Both are designed against human sequence. For amplicon assays the relevant measurement is old and precise: in allele-specific polymerase chain reaction (PCR), the copying reaction tuned to amplify one version of a sequence and not another, a single mismatch at a primer's 3'-terminal T, G or C — the last letter at the end the enzyme extends from — costs 40- to 100-fold of the product, a mismatch one base in costs 8- to 20-fold, and a 3'-terminal A discriminates poorly ³⁰. How those penalties combine across two primers and several mismatches was not measured, and the figures do not describe hybridization capture. The inference that a human hotspot assay would not amplify fluke DNA follows from this measurement and from the sequence divergence described later on this page; it is a strong inference and it is not a measurement, because the experiment has not been published.

Depth and scale differ enormously across the family. A hotspot panel of 50 genes validated on 250 formalin-fixed specimens reached 99.1% sensitivity with 98.5% concordance against older methods ¹⁰. A 341-gene capture assay detected known variants across 47 exons — the protein-coding stretches — of 19 genes with a detection limit near 2% of reads for hotspots ³. A 287-gene assay reported 95% to 99% sensitivity with positive predictive value above 99%, and found clinically actionable alterations in 76% of 2,221 cases, three times what the diagnostic tests of the day were finding ¹¹. A 523-gene assay adds fusions (two genes joined by a rearrangement) and splice variants in 55 genes, copy number, tumor mutational burden and microsatellite instability, a stutter in short repeated sequences that signals a broken DNA mismatch-repair system, from paired DNA and RNA libraries ¹². Whole-exome and whole-genome sequencing remove the gene list — exome capture was demonstrated on twelve human exomes and more than 300 megabases of coding sequence in 2009 ⁵¹ — but they do not remove the reference genome, which is the filter that matters for this season.

What the scale buys, in human terms, is large. Prospective sequencing of more than 10,000 patients with advanced cancer enrolled 11% on trials chosen to match a mutation in their tumor ^66,67. Whole-genome sequencing of 13,880 solid tumors across 33 cancer types in the United Kingdom found small variants (single-letter changes and short insertions or deletions) in 94% of glioblastomas, copy-number aberrations in at least one recommended gene in 58%, actionable structural variants (large rearrangements) in 13% of sarcomas, and homologous recombination deficiency — a failure of one DNA-repair pathway that makes tumors sensitive to platinum drugs and PARP inhibitors — in 40% of high-grade serous ovarian cancers, 30% of it from inherited variants ³³. None of this is in doubt, and none of it is a search for anything non-human.

How the result is reported, and who decides what to order

A karyotype is reported in a notation that names the chromosome, the arm, the region and the band — which is why the band stage matters, and why a laboratory states whether a preparation was read at the 400-, 550- or 800- to 850-band level ⁷. A panel is reported as a list of variants with their positions in the reference sequence and the fraction of reads carrying each one, plus whatever composite measures the assay produces ¹². Neither standard report has a line for what was found and not recognized; there is no field for unassigned sequence. The unassigned reads are not destroyed, though: they sit in the alignment file as unmapped reads, and one laboratory has proposed reporting the Epstein-Barr virus read count from its 400-gene panel as a routine result, with ten or more reads as the threshold that matched tissue staining ³⁶.

What gets ordered is set by guidelines, and the guidelines are explicit. The 2022 European LeukemiaNet recommendations for acute myeloid leukemia build a revised genetic risk classification on cytogenetics together with molecular markers ³², and the American technical standards for cytogenomic studies — chromosome and genome-wide testing — of neoplastic blood, marrow and lymph nodes were updated in 2024, superseding the 2016 edition ^31,72. For advanced solid tumors, the European Society for Medical Oncology recommends tumor sequencing in non-squamous non-small-cell lung, prostate, colorectal, biliary and ovarian cancer, and in the 2024 update extended that to breast cancer and to rare tumors including gastrointestinal stromal tumors, sarcoma, thyroid cancer and cancer of unknown primary ⁴; a corrigendum, a published correction, to that report was published in 2025 ⁷³. The American Society of Clinical Oncology's 2022 provisional clinical opinion says patients with metastatic or advanced cancer should have genomic sequencing in a certified laboratory whenever a regulator-approved biomarker exists for their disease, with a multigene panel when more than one biomarker-linked therapy is approved, and that site-agnostic approvals for high mutation burden, mismatch-repair deficiency and NTRK fusions justify testing in all solid tumors ⁴². For colorectal cancer specifically, a four-society American guideline built from more than 4,000 articles issued 21 statements and recommends mutational testing of genes in the epidermal growth factor receptor pathway ³⁴. For plasma DNA, the 2026 American Society of Clinical Oncology guideline, built from 54 meta-analyses and 22 study reports including seven randomized trials, says ctDNA testing for tumor alterations may be used when tissue testing is challenging, unsafe or too slow or when a drug's approval allows it, that a negative, inconclusive or clinically inconsistent result should be confirmed in tissue, and that ctDNA fraction or concentration is not recommended as a surrogate measure of disease ⁷⁰.

Two honest notes about those documents. First, the colorectal guideline's author block includes two members whose listed affiliations are commercial molecular-diagnostics companies, Castle Biosciences and Biocept ³⁴; the individual disclosure statements of the European panels are not carried in the records used here, and should not be presented as absent. Second, guidelines are consensus about which questions are worth asking. The 2024 European document is a list of cancers in which gene sequencing changes management ⁴. It is not, and does not claim to be, a judgment about what else might be in a specimen.

The reference genome problem

Here is the step that decides this whole question. A sequencing run produces hundreds of millions of short reads; software aligns each one to the human reference genome; and what will not align is written to a side file, flagged, or thrown away. The reference has been improving for twenty years and is still a construction: a 2017 assessment of GRCh38, the 2013 build of the human reference that most clinical laboratories still align to, documented the resolution of roughly a thousand issues and the first sequence representation of the centromeres, the pinched waists of the chromosomes ⁷⁴, an independent survey found that about 5% of the reference length was still unresolved gaps and closed 132 of 783 of them with 2.2 megabases of new sequence ¹⁴, and the first gapless assembly of a human genome — 3.055 billion base pairs, complete for every chromosome except Y, with nearly 200 megabases of new sequence and 1,956 gene predictions — appeared only in 2022 ¹³. The complete genome changes what the aligner does with reads: against 3,202 short-read samples it improved mapping and variant calling across the board, found hundreds of thousands of variants per sample in previously unresolved regions, and eliminated tens of thousands of spurious variants per sample, cutting false positives in 269 medically relevant genes by up to twelvefold ⁶¹. In 2023 a first draft human pangenome — 47 complete genomes from genetically diverse people, adding 119 million base pairs and 1,115 gene duplications that GRCh38 lacks — reduced small-variant errors by 34% and doubled the structural variants detected per genome copy ¹⁵. Most clinical pipelines still align to GRCh38, and the moral runs in one direction: a read that fails to match the reference is at least as likely to be human sequence the reference lacks as anything foreign.

Reads that fail to align are therefore a mixture of three things: real non-human sequence, human sequence the aligner could not place, and laboratory artifact. The proportions are not favorable to the exciting interpretation. A survey of 150 sequencing data files, whole-genome and exome, from the 1000 Genomes Project found that on average 0.13% of reads resembled non-human genomes, that sequencing centers carried their own characteristic contaminating signatures like time stamps, and that many unmapped reads looked foreign only because human sequence resembles stretches of the mouse and tobacco assemblies ¹⁶. Two later surveys explain why: human DNA has contaminated 2,250 bacterial and archaeal (a second domain of single-celled microbes) genome assemblies, generating 3,437 spurious protein entries in widely used databases ⁷⁵, and a systematic search found more than 2.1 million contaminated entries in RefSeq, plus 114,035 in GenBank, the public sequence databases every taxonomic search relies on ⁷⁶.

Pipelines that genuinely hunt for pathogens are built the other way around. They subtract the host first and carefully: digital transcriptome subtraction uncovered a tumor virus in 2008 ⁵⁷, PathSeq packaged the approach as software in 2011 ⁷⁷, and a 2026 tumor pipeline that ends with three aligner-and-reference pairs removes more than 99.9% of human reads while, in simulations, preserving more than 99.7% of true bacterial signal, precisely because in tumor data the non-human reads are a small minority ⁷⁸. Where such work has been done rigorously it has produced real findings: a polyomavirus in Merkel cell carcinoma, by subtracting human transcripts ⁵⁷; Fusobacterium nucleatum over-represented in colorectal tumor tissue, confirmed by quantitative polymerase chain reaction (PCR), a counting version of the DNA-copying test, in 99 subjects ⁷⁹, later shown to persist in distal metastases and in xenografts, human tumors grown in mice, where an antibiotic reduced tumor growth ⁸⁰; and viruses detected in 382 genome and 68 RNA (transcriptome) datasets across 2,658 cancers in the pan-cancer whole-genome analysis ^81,82. None of those discoveries came out of a clinical cancer panel. Each needed a purpose-built pipeline and independent confirmation.

The contamination risks that are measured and real in clinical oncology are duller than parasite DNA. DNA from extraction kits and reagents is ubiquitous, varies between kits and between batches of the same kit, and dominates low-biomass samples — which is why the right question to ask a laboratory is always what the negative control showed ⁸³. On sequencers of one common design, reads got mislabeled between pooled patient samples at 0.2% to 6% per run, a problem characterized in 2018 along with its remedy, non-redundant dual indexing, giving every sample two unique barcodes so that a swapped read can be recognized and discarded ⁸⁴. And the field has a worked example of a low-biomass microbiome that turned out not to exist: careful work found no evidence of bacteria in most human placentas, with almost every signal traceable to acquisition during labor or to reagent contamination ⁸⁵.

What was claimed from tumor sequencing data, and what collapsed

In 2020 two prominent papers put non-human sequence inside human tumors. One analyzed 1,526 tumors and their adjacent normal tissue across seven cancer types and reported a distinct bacterial composition for each type, mostly intracellular, while stating plainly that the low biomass involved makes such work difficult ⁶³. The other re-examined whole-genome and whole-transcriptome (all-RNA) data from 18,116 samples across 33 cancer types in The Cancer Genome Atlas, reported unique microbial signatures within and between most cancers, and claimed to separate 69 cancer-free people from 100 samples from patients with prostate cancer, lung cancer or melanoma using plasma microbial nucleic acids alone, after decontamination that discarded up to 92.3% of the sequence data ⁶⁰. A third strand concerned fungi, which are eukaryotes like parasites. In 2022 two Cell papers reported fungal DNA at low abundance across 17,401 tissue, blood and plasma samples from 35 cancer types in four cohorts, with compositions that differed by cancer type ⁸⁶, and tumor-associated fungi in gastrointestinal and lung cancers, with Candida predicting metastasis in colon cancer and decreased survival across gut sites ⁸⁷. The first shares authors and methods with the retracted paper and lists affiliations with the same company, and the 2025 re-analysis compared its counts with earlier claims about bacteria, viruses and fungi ⁶⁴.

The second paper was retracted by Nature in July 2024 ⁶². The authors did not accept the criticism. In February 2024 they published a re-analysis in Oncogene reporting that batch correction gave equivalent predictions on raw and corrected data, that about 1% of genomes in their original databases carried human contamination, that the extra human reads the critics found came from using a newer human reference, and that cancer type-specific microbiomes persisted when they re-ran everything with the critics' own pipeline ⁶⁵; that paper carries its own published correction ⁸⁸, and the journal that published it has not retracted it. Several of its authors, including the first, list affiliations with Micronoma, a company set up to sell microbial cancer diagnostics ⁶⁵. The re-analysis that preceded the retraction is worth reading carefully, because its failure mode runs opposite to the hypothesis this season is testing. Re-aligning the unmapped reads of 1,255 TCGA samples from three cancer types (tumor, normal tissue and blood) to the complete CHM13 human genome recovered an average of 1.4 to 3.3 million additional human reads per sample; the original search database contained 59,974 microbial genomes including contaminated drafts, and no human genome and no laboratory vectors, the engineered carrier DNA used in cloning, so those human reads were matched to bacteria instead. In one bladder tumor, two genera (groups of related species) reported with 327,985 and 20,673 reads yielded one read and zero reads on re-analysis; when the original-style search was re-run against 2016 draft genomes without the human genome, 98.1% and 98.9% of the reads assigned to those two genera aligned to human DNA. Across three cancer types, 92.9% to 98.5% of sample-genus counts of 10 or more were at least ten times too high ¹⁷.

The second error was worse. The re-analysts built a matrix of 16,567 samples and 170 genera in which every raw value was zero, filled it with the published normalized values (66 of those genera had them), kept the 12,803 primary tumors, ran the original code, and obtained classifiers with a median sensitivity of 0.94 (the share of cancers correctly called) and a median specificity of 0.9 (the share of non-cancers correctly called) — with 14 of 32 models matching or beating the original performance, from data that contained no information at all before normalization ¹⁷. A classifier that works on zeros is not a discovery about cancer. It is a description of a transformation.

This is the discipline the subject requires, and it cuts against the owner's hypothesis rather than for it. The best-known attempt to read non-human sequence out of the largest human tumor archive failed because human reads were misclassified as microbial. In the best-known research pipeline, the documented failure ran toward seeing foreign sequence that was not there ¹⁷; clinical oncology pipelines do not classify unmapped reads at all, so they have no measured bias in either direction. Any claim that molecular cancer testing is quietly picking up non-human material has to clear that bar first, and it has to say out loud that more than a dozen downstream studies built on the retracted data are affected as well ¹⁷. The re-analysis has since been extended to all 5,734 whole-genome sequencing datasets then available from The Cancer Genome Atlas, across 25 cancer types, with updated methods and databases: the presence of microbes was far smaller than had been reported, many reported species may not be present at all, and the authors released read counts for bacteria, viruses, archaea and fungi in every sample so that others are not misled ⁶⁴. In 2026 a group including the lead re-analyst set out the quality controls and validation steps a report of microbes in a tumor should meet, comparing the problem to contamination in ancient-DNA work ⁸⁹.

Parasite counterparts of cancer genes, example by example

The owner asked whether parasites carry the genes our cancer tests look for. For worms the answer is largely yes, and it is more interesting than it first appears. Two genes descended from the same ancestral gene in two species are orthologs; two that arose by duplication within a lineage are paralogs, and the distinction governs what you may infer about function ⁹². The p53 family is the deep case: an ancestor of human p53, p63 and p73 is first detected in sea anemones, is present in almost all invertebrates, and the family has preserved its structure and activities for over a billion years ⁹³. So a recognizable p53 in a flatworm is expected, not surprising.

It has now been characterized. A parasitic flatworm carries both a true ortholog of planarian p53 and human TP53, governing stem cell maintenance and skin production, and a second, parasite-specific paralog required for the normal response to DNA-damaging stress, which the authors attribute to convergent evolution ²². In December 2025 the same group reported — the work had circulated as a preprint since May 2024 ⁹⁴ — that the schistosome p53 ortholog controls maintenance of the worm's skin, that a cyclin-dependent kinase inhibitor homolog causes hyperproliferation when knocked down, and that knocking down both produces tumor-like growths — identifying both as tumor suppressors in the parasite. These CDK-inhibitor homologs are ubiquitous in parasitic flatworms and absent from their free-living ancestors, which the authors read as a sign of horizontal gene transfer ²³. A 2026 commentary names the two genes cki and p53-1 and reads the transfer as support for a new hypothesis about how parasitic flatworms arose ⁹⁵. Note the direction of that finding: the worm has cancer genes because the worm needs tumor suppression, not because it is seeding ours.

The other examples check out individually, and each carries a caution. Schistosome Ras is a single-copy gene whose protein is 81% identical and 92% similar to K-Ras, sits in the layer beneath the worm's skin, carries a fatty farnesyl tag that anchors it to the cell membrane, and had that tagging blocked by a farnesyl transferase inhibitor, a drug class once tried against human Ras, at concentrations comparable to those needed for K-Ras ²⁴ — but those are protein-level figures, and protein identity permits large divergence in the underlying DNA because several triplets code for the same amino acid. The fox tapeworm carries Ras and Raf orthologs: EmRas is 79% identical to human Ras and functionally replaced the yeast gene in a growth assay, and the emraf gene shares four conserved exon-intron boundaries — the points where coding stretches meet the stretches cut out of the message — with the human gene for Raf-1 ²⁵. A blood fluke EGFR homolog, SER, was cloned in 1992 by homology to the tyrosine kinase domain — the part of the protein that attaches phosphate to switch other proteins on — of erbB, the gene family EGFR belongs to ²⁶. That is the same domain clinical EGFR mutation tests read, since the activating mutations are deletions and substitutions in the kinase domain, clustered around its ATP-binding pocket ⁹⁶, so this example does not show divergence at the tested positions, and the nucleotide identity across those positions has not been reported.

The list continues past the famous oncogenes. The blood fluke has a kinase that is a hybrid of the human Src and Abl enzymes, whose inhibition profile sits between Src and Abl inhibitors ²⁷, and imatinib — the drug that came out of the Philadelphia chromosome story — damages adult worms and kills them in culture ⁹⁷. Tapeworms have a TOR pathway: Echinococcus granulosus larvae express TORC1, the protein complex at the center of the TOR growth switch, and respond to rapamycin, the drug that blocks it ²⁸, and tacrolimus — which the authors grouped with inhibitors of mTOR, the human version of TOR, although it is pharmacologically a calcineurin inhibitor — given orally at 4 mg/kg per day reduced cyst weight and number in mice ⁹⁸. And the liver fluke secretes a granulin, Ov-GRN-1, that stimulated murine fibroblasts at nanomolar concentrations and is the main growth factor in the fluke's secretions ²⁹; at low nanomolar levels it drives human cholangiocytes, the cells lining the bile ducts, to proliferate ⁹⁹. Every one of those is animal or laboratory evidence about the parasite or about a drug, and none of it is evidence that a human test can see the gene. The newest counts are whole kinomes — a kinome is an organism's full set of protein kinases, the enzymes that switch other proteins on by attaching a phosphate. Two annotations of the sheep liver fluke Fasciola hepatica, a relative of the carcinogenic flukes that is not itself a Group 1 agent, found 245 and 271 kinases, roughly 1.6% to 2.1% of its proteins, in family proportions like those of other parasitic flatworms ^100,101. By orthology to human sequence, a quarter of the 110 kinases that could be assigned to pathways fell on cancer pathways, and two human kinase inhibitors, vandetanib and ruboxistaurin, killed immature flukes in culture at 50 micromolar ¹⁰⁰. Silencing the fluke's polo-like kinase 1, a cell-division kinase, cut growth and stem-cell proliferation in juvenile worms, and a human drug against that kinase, BI 2536, did the same in a dose-dependent way; three of that paper's authors work for an animal-health company ¹⁰¹. All of this is in vitro, and none of it says whether a human assay would read the fluke's gene.

Two examples the brief asked for do not check out, and the page says so. No MYC ortholog and no BRAF ortholog in a helminth was found in these searches; what is verified is p53 with its paralog, a CDK inhibitor, Ras, Raf, an EGFR homolog, Src- and Abl-type kinases, TOR and granulin. And for at least one major single-celled parasite the answer is no: an exhaustive analysis of the malaria parasite's 65 protein kinases found none in the tyrosine kinase group, alongside a family of twenty kinases restricted to the Apicomplexa, the branch of single-celled parasites that includes malaria and Toxoplasma ¹⁰². An organism with no tyrosine-kinase-group kinases has no EGFR, ALK or KIT — the receptor kinases that lung, lymphoma and stromal-tumor panels read — for a panel to find. The homology argument is strongest for worms, which are animals; for Plasmodium it fails, and other protozoa were not examined here.

The tests that would see a parasite, and the one that did

If the question is whether something non-human is in a specimen, there are assays for that, and they are not these assays. The method with the best claim is broad-range amplification: primers aimed at a stretch of DNA shared by all nucleated life, which can return an organism nobody thought to look for. In 1996 that approach amplified 357 base pairs of ribosomal DNA directly from the tissue of a man whose abdominal mass had been taken for a neoplasm and identified a previously uncharacterized tapeworm ⁴⁵. In 2015 a polymerase chain reaction targeting eukaryotes identified Hymenolepis nana DNA in the proliferating cells of a man with HIV, and immunohistochemistry and probe hybridization then put worm material physically inside those cells ⁴⁴. Neither identification came from a cancer gene panel; both used assays designed to find non-human DNA, and whether a panel would have registered anything has never been tested.

Clinical metagenomic sequencing — reading all the DNA in a sample, from whatever organism, and sorting it by species — is now offered in infection medicine, and the case literature shows it finding parasites when it is pointed at them. A validated plasma assay covers 1,250 organisms including eukaryotic parasites and agreed with blood culture in 93.7% of 350 patients with a sepsis alert, identifying a cause more often than all the other microbiological testing combined; it is a commercial laboratory's test, which belongs in any account of it ¹⁰⁴. In tissue the recent case literature is explicit: 6,156 reads matching Echinococcus multilocularis in a lung puncture sample ¹⁰⁵, Schistosoma japonicum identified in brain tissue from a three-year-old after surgery ¹⁰⁶, and Spirometra identified in cerebrospinal fluid and then confirmed in a brain lesion biopsy ¹⁰⁷. These are case reports, not accuracy studies, and they are evidence of what the technology can do rather than how often it is right. There is also an accuracy study. Over seven years one university laboratory ran its cerebrospinal-fluid metagenomic test on 4,828 samples and detected 797 organisms in 697 of them (14.4%): 363 DNA viruses, 211 RNA viruses, 132 bacteria, 68 fungi and 23 parasites (2.9%). Against clinical diagnoses in a subset of 1,164 samples the test's sensitivity was 63.1% and its specificity 99.6%, and 48 of 220 infectious diagnoses (21.8%) were made by metagenomics alone. Several authors work for the company that now sells the test ¹⁰⁸.

What is not available is the one measurement that would settle the episode's central question. No published study found in these searches has run a clinical cancer hotspot assay, a companion diagnostic — the test approved alongside a drug to choose its patients — or a hybridization-capture panel against purified helminth or protozoan DNA and reported whether anything amplified. No audit of what accredited solid-tumor laboratories do with their unmapped reads was found. What was found is narrower and real: at least two academic laboratories have taken the unmapped or off-target reads from their clinical cancer panels and searched them for non-human sequence, one in 21 gliomas on a 151-gene panel ³⁵, one in 5,234 blood-cancer samples on a 400-gene panel ³⁶, and both found viruses. Neither looked for worms or protozoa. The spiking experiment is publishable: graded quantities of fluke and tapeworm DNA into human tumor DNA, through a 500-gene panel and a hotspot assay, and report cross-reactivity. A second experiment is cheaper still: search the archived off-target reads of an existing panel series for helminth and protozoan sequence, as was done for Epstein-Barr virus ³⁶. Until someone does it, the correct sentence on air is that these searches found no published experiment of this kind.

What was believed and is not

Four beliefs in this subject have been corrected, and only one of the corrections runs in the direction the subject invites. Two are from the founding years of cytogenetics, when a number and then a mechanism were both read wrong for years at a time. One is about the karyotype's own sensitivity, which turned out to be lower than the field had assumed. One is the retraction that governs any claim about non-human sequence in tumor data. A fifth entry is not a corrected belief but a measurement the parasite-cancer literature has already made, about which p53 is elevated in a parasite-associated tumor and in whose tissue it was measured.

They are listed with the date the correction landed, because in this show a belief and its correction travel together. Three of them share a shape worth naming: an accepted answer shaped what careful people saw next, whether that answer was forty-eight chromosomes, a deleted chromosome arm, or a microbial signature that a normalization step had written into the data. The fourth is a correction of scope rather than of fact: the test was weaker than believed. The fifth is a measurement rather than a correction, and it is here because it answers the question directly: the molecule measured belonged to the patient, and nobody tested for the worm's.

What is still unknown

Nobody has tested a clinical cancer panel against purified parasite DNA. That is the central empirical hole, and it is a hole in both directions: there is no measurement showing that a human hotspot assay cannot amplify fluke template, the DNA being copied,, and none showing that it can. The argument on this page is built from a measured penalty for primer mismatch, 40- to 100-fold for a single wrong T, G or C at the 3' end ³⁰, and from protein-level divergence figures such as the schistosome Ras at 81% identity to K-Ras ²⁴. That is a strong inference and it is not a measurement. The nucleotide-level identity, which is the number that actually governs whether a primer anneals, is not published for these genes.

Nobody has audited what clinical laboratories do with unmapped reads. Whether they are retained in the alignment file, written elsewhere, or discarded; for how long; and whether any taxonomic classification step exists in a validated oncology pipeline — none of this could be established for laboratories in general. Two laboratories have published what they did with those reads, and in both cases the reads were kept and could be searched ^35,36. The adjacent evidence is indirect but real: pathogen pipelines subtract the host first by design ⁷⁷, and one 2026 pipeline that recovers microbial reads from tumor whole-genome data rigorously used three human references and a purpose-built tool, trading some ambiguous microbial signal for stricter host removal ⁷⁸. Absence from a search is not proof of absence, and this page says so in those words. No published search of tumor sequencing archives for helminth or protozoan sequence was found. The 2025 re-analysis of 5,734 Cancer Genome Atlas genomes counted bacteria, viruses, archaea and fungi ⁶⁴; parasites were not among its targets.

The Hymenolepis case remains a single patient, eleven years on. No replication, no second case, no series; the only adjacent records found were the 2016 correspondence and the authors' reply ^44,58,59. The nearest established principle is adjacent rather than identical: eleven transmissible cancer lineages are known across dogs, Tasmanian devils and bivalves, and some of them cross species ¹¹⁰ — but those are cancers passed from one animal to another, whereas the Hymenolepis case was a tapeworm's own cells proliferating inside a human host, which no lineage on that list does. The count is still the discipline: eleven lineages in all of biology, and exactly one published instance of a parasite's cells behaving like a cancer in a person. A related fatal tapeworm, Sparganum proliferum, has had its genome sequenced, and its proliferation in human tissue is attributed by the authors to unusual coordination of the extracellular matrix, the scaffolding between cells, and to loss of sexual maturity ¹¹¹.

Two smaller gaps deserve a sentence each. There is no verified MYC or BRAF ortholog in a helminth in this literature, so the page does not claim one, and the right phrasing on air is that these searches did not find it rather than that it does not exist. And the index-swapping figure of 0.2% to 6% comes from named 2018 instruments, with the paper's own remedy, non-redundant dual indexing ⁸⁴; how widely clinical laboratories use it today was not established here, so quoting 6% as today's clinical rate would be misleading. Cross-sample bleed is real, it is measured, and it moves reads between human patients rather than between species.

How this connects to parasites and cancer, at its true strength

Start with what is not in dispute. Three helminths are International Agency for Research on Cancer Group 1 agents, the tier reserved for sufficient evidence of carcinogenicity in humans: Opisthorchis viverrini, Clonorchis sinensis and Schistosoma haematobium ¹¹². These parasites and their relatives have real, sequenced genomes — the blood fluke S. mansoni, a relative of the Group 1 S. haematobium but not itself a Group 1 agent, at 363 megabases and at least 11,809 genes with more than 300 proteases, enzymes that cut proteins ¹¹³, C. sinensis now at chromosome level, affecting about 35 million people ¹¹⁴ — and real chromosome counts: eight pairs in schistosomes, with sex chromosomes arranged ZZ in males and ZW in females, the reverse of our XX and XY ¹⁸, 2n=12 in O. viverrini confirmed by two independent groups, one of which prepared its chromosomes from the testes of adult worms ^19,20, 2n=14 in C. sinensis ¹⁹, and 2n=12 in the dwarf tapeworm, recorded in the literature under the genus Vampirolepis ²¹. Twelve chromosomes against a human 46 is not a subtle confusion. If anyone did karyotype worm material, the difference would be unmissable.

The mechanism by which these parasites actually cause cancer is known in outline, and it is host-side. Fluke granulin pushes human bile-duct cells to divide, shown in vitro at low nanomolar concentrations ⁹⁹, confirmed by knockdown that cut proliferation of a cholangiocarcinoma line by 92% ⁹⁹, and carried into an animal by programmed gene editing, where gene-edited worms with depleted granulin still colonized the bile ducts but caused less overgrowth of the bile-duct lining (hyperplasia) and less scarring (fibrosis) ¹⁰³. A 2024 review by the granulin's discoverers states the current mechanism as secreted growth factors, digestive enzymes and extracellular vesicles — tiny membrane packets the worm sheds — acting on bile-duct cells together with inflammation and repeated wounding where the worm feeds and grazes on the lining ¹¹⁵. The same group is developing fragments of the fluke granulin as wound-healing agents, which is worth knowing when weighing its claims ¹¹⁶. Schistosome egg antigens switch on Wnt and β-catenin signaling, a growth pathway, and c-Jun, a proto-oncogene — a normal gene that drives cancer when overactive — in gut lining cells, with the same markers found in colon biopsies from infected patients ¹¹⁷. Fluke infection also reshapes the biliary and intestinal microbial community in an animal model that reliably produces cholangiocarcinoma ¹¹⁸. And the tumors themselves carry the signature: of 209 bile duct cancers, 15 sequenced across the protein-coding exome and 194 screened for the recurrent mutations, fluke-related and non-fluke-related disease showed different mutation patterns, with TP53 mutations more frequent in the fluke-related group ¹¹⁹, a pattern confirmed in an integrated analysis of 489 cases from ten countries ¹²⁰. These are the patient's genes being driven, in the patient's cells. That is carcinogenesis, not a testing artifact.

Now the narrow claim, stated at its real strength. Parasite nucleic acid does circulate in infected people, and in infected mice it reaches the heart: cell-free Schistosoma DNA was found in the blood of 6.98% of 186 Egyptian patients with infective endocarditis, and in infected mice in 95% of blood samples and 45% of cardiac tissue samples — and the authors' call to investigate whether such DNA can trigger inflammation, abnormal proliferation or genome integration is a call for research, not a finding ¹²¹. So parasite DNA may be present in tissue a pathologist might biopsy, though this study showed that only in mice. It was found by a species-specific assay designed to find it. And the molecular consequences that have been documented in the parasite-associated cancers are host-side changes measured in host tissue, including the methylation — chemical tags that switch genes off — of host genes detectable in urine in S. haematobium bladder disease ¹²².

The conclusion this page supports is narrow and worth saying exactly. A karyotype would usually not see a parasite because most specimens hold no dividing parasite cells, and where they do — as in the Hymenolepis case ⁴⁴ — a worm's 2n=12 set would look nothing like ours ²¹. A gene panel's report would not show a parasite because it reads a list of human genes with human-designed probes, and the reference genome step sets aside whatever does not match. Set aside is not thrown away: the same off-target reads have yielded viral genomes when someone searched them ^35,36,90, and nobody has searched them for a worm. Neither absence is evidence that nothing non-human is there; both are evidence that nobody asked. The assay that would ask exists and is not ordered in a tumor workup. That is a real gap in medical practice, and it is a smaller and more defensible claim than the one the gap invites — and it would take one spiking experiment and one laboratory audit to turn it from an argument into a measurement.

Where it connects

Sources

122 sources, numbered as they are cited. Every one was checked against PubMed or its publisher before it was cited here; the note under each says what it shows and what it does not.

1. 1
   Caspersson T, Zech L, Johansson C. Differential binding of alkylating fluorochromes in human chromosomes.Exp Cell Res · 1970 · 60(3):315-9doi:10.1016/0014-4827(70)90523-9 · PMID 5422961

   The first fluorescent banding of human chromosomes. The PubMed record has no abstract; only the title and indexing support the claim made here.

2. 2
   Seabright M. A rapid banding technique for human chromosomes.Lancet · 1971 · 2(7731):971-2doi:10.1016/s0140-6736(71)90287-x · PMID 4107917

   The trypsin method that made banded karyotypes routine. No abstract in the record; cited for the method and its date only.

3. 3
   Cheng DT, Mitchell TN, Zehir A, et al. Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT): A Hybridization Capture-Based Next-Generation Sequencing Clinical Assay for Solid Tumor Molecular Oncology.J Mol Diagn · 2015 · 17(3):251-64doi:10.1016/j.jmoldx.2014.12.006 · PMID 25801821

   A 341-gene capture panel for fixed tumors, validated on 284 samples with known variants in 47 exons of 19 genes; detection limit about 2% for hotspots and 5% elsewhere. An institutional assay validation.

4. 4
   Mosele MF, Westphalen CB, Stenzinger A, et al. Recommendations for the use of next-generation sequencing (NGS) for patients with advanced cancer in 2024: a report from the ESMO Precision Medicine Working Group.Ann Oncol · 2024 · 35(7):588-606doi:10.1016/j.annonc.2024.04.005 · PMID 38834388

   The governing European recommendation for solid tumors, written entirely in the language of genes and panels; karyotyping does not appear. Expert consensus, with individual disclosures not in the record.

5. 5
   Gartler SM. The chromosome number in humans: a brief history.Nat Rev Genet · 2006 · 7(8):655-60doi:10.1038/nrg1917 · PMID 16847465

   Historical analysis of the 1956 correction of the human chromosome number; it is a historian's account, not primary cytogenetic data.

6. 6
   Harper PS. The discovery of the human chromosome number in Lund, 1955-1956.Hum Genet · 2006 · 119(1-2):226-32doi:10.1007/s00439-005-0121-x · PMID 16463025

   Dates the determination of 46 to December 1955 with publication in April 1956, ending more than thirty years of 48, and names the technical advances required.

7. 7
   Magenis RE, Barton SJ. Delineation of human prometaphase paracentromeric regions using sequential GTG- and C-banding.Cytogenet Cell Genet · 1987 · 45(3-4):132-40doi:10.1159/000132444 · PMID 3502699

   Source for the 400, 550 and 800-850 band stages as working levels of karyotype resolution; the paper itself is about paracentromeric regions.

8. 8
   Kakazu N, Bar-Am I, Hada S, et al. A new chromosome banding technique, spectral color banding (SCAN), for full characterization of chromosomal abnormalities.Genes Chromosomes Cancer · 2003 · 37(4):412-6doi:10.1002/gcc.10229 · PMID 12800153

   States that donor metaphases ranged from 250 to 850 bands per haploid set; also an example of G-banding failing to characterize a lymphoma rearrangement that a multicolor method resolved.

9. 9
   Haley L, Parimi V, Jiang L, et al. Diagnostic Utility of Gene Fusion Panel to Detect Gene Fusions in Fresh and Formalin-Fixed, Paraffin-Embedded Cancer Specimens.J Mol Diagn · 2021 · 23(10):1343-1358doi:10.1016/j.jmoldx.2021.07.015 · PMID 34358677

   The cleanest evidence for the fresh-versus-fixed split inside one laboratory: 73 fresh leukemia and lymphoma specimens with conventional karyotyping as a comparator, against 141 formalin-fixed solid tumors tested by nucleic acid.

10. 10
    de Leng WWJ, Gadellaa-van Hooijdonk CG, Barendregt-Smouter FAS, et al. Targeted Next Generation Sequencing as a Reliable Diagnostic Assay for the Detection of Somatic Mutations in Tumours Using Minimal DNA Amounts from Formalin Fixed Paraffin Embedded Material.PLoS One · 2016 · 11(2):e0149405doi:10.1371/journal.pone.0149405 · PMID 26919633

    A 50-gene hotspot panel on 250 fixed specimens: 99.1% sensitivity and 98.5% concordance with conventional sequencing. One commercial co-author affiliation is listed.

11. 11
    Frampton GM, Fichtenholtz A, Otto GA, et al. Development and validation of a clinical cancer genomic profiling test based on massively parallel DNA sequencing.Nat Biotechnol · 2013 · 31(11):1023-31doi:10.1038/nbt.2696 · PMID 24142049

    A 287-gene fixed-tissue assay with 95-99% sensitivity and positive predictive value above 99%, reporting actionable alterations in 76% of 2,221 cases. Authored largely by the company selling the test.

12. 12
    Giacò L, Palluzzi F, Guido D, et al. A Computational Framework for Comprehensive Genomic Profiling in Solid Cancers: The Analytical Performance of a High-Throughput Assay for Small and Copy Number Variants.Cancers (Basel) · 2022 · 14(24):6152doi:10.3390/cancers14246152 · PMID 36551638

    Source for a 523-gene panel that also reports fusions in 55 genes, copy number, mutational burden and microsatellite instability; an internal validation on 71 DNA and 64 RNA samples of a named commercial assay.

13. 13
    Nurk S, Koren S, Rhie A, et al. The complete sequence of a human genome.Science · 2022 · 376(6588):44-53doi:10.1126/science.abj6987 · PMID 35357919

    The first gapless human genome: 3.055 billion base pairs, the remaining 8% addressed, nearly 200 megabases of new sequence and 1,956 gene predictions. One cell line, and no Y chromosome in this assembly.

14. 14
    Zhao T, Duan Z, Genchev GZ, Lu H. Closing Human Reference Genome Gaps: Identifying and Characterizing Gap-Closing Sequences.G3 (Bethesda) · 2020 · 10(8):2801-2809doi:10.1534/g3.120.401280 · PMID 32532800

    States that hundreds of reference gaps remained, about 5% of total sequence length, and closes 132 of 783 with 1,125 sequences and 2.2 megabases; most of the new sequence is repetitive. A computational study.

15. 15
    Liao WW, Asri M, Ebler J, et al. A draft human pangenome reference.Nature · 2023 · 617(7960):312-324doi:10.1038/s41586-023-05896-x · PMID 37165242

    47 phased diploid assemblies; 119 million base pairs of polymorphic sequence and 1,115 gene duplications added relative to GRCh38; 34% fewer small-variant errors and 104% more structural variants per haplotype than GRCh38 workflows. Co-authors at Google and Dovetail Genomics.

16. 16
    Tae H, Karunasena E, Bavarva JH, et al. Large scale comparison of non-human sequences in human sequencing data.Genomics · 2014 · 104(6 Pt B):453-8doi:10.1016/j.ygeno.2014.08.009 · PMID 25173571

    On average 0.13% of reads in 150 sequencing data files (whole-genome and exome) from the 1000 Genomes Project resembled non-human genomes, sequencing centers carried characteristic contaminating signatures, and many unmapped reads only looked foreign because human sequence resembles the mouse and tobacco assemblies.

17. 17
    Gihawi A, Ge Y, Lu J, et al. Major data analysis errors invalidate cancer microbiome findings.mBio · 2023 · 14(5):e0160723doi:10.1128/mbio.01607-23 · PMID 37811944

    Primary computational re-analysis, not a review: 1.4-3.3 million human reads left per sample, counts at least tenfold too high in 92.9-98.5% of entries, 98.1% and 98.9% of two genera's reads actually human, and near-perfect classifiers rebuilt from an all-zero matrix. The authors note the classification tool was developed in one of their own labs.

18. 18
    Stitz M, Chaparro C, Lu Z, et al. Satellite-Like W-Elements: Repetitive, Transcribed, and Putative Mobile Genetic Factors with Potential Roles for Biology and Evolution of Schistosoma mansoni.Genome Biol Evol · 2021 · 13(10):evab204doi:10.1093/gbe/evab204 · PMID 34469545

    States that the schistosome karyotype comprises eight chromosome pairs, with homogametic ZZ males and heterogametic ZW females. The paper's own subject is repetitive DNA.

19. 19
    Zadesenets KS, Katokhin AV, Mordvinov VA, Rubtsov NB. Comparative cytogenetics of opisthorchid species (Trematoda, Opisthorchiidae).Parasitol Int · 2012 · 61(1):87-9doi:10.1016/j.parint.2011.07.006 · PMID 21798365

    2n=14 for Clonorchis sinensis, Opisthorchis felineus and two Metorchis species, and 2n=12 for Opisthorchis viverrini. The Clonorchis count rests on this group alone.

20. 20
    Kaewkong W, Choochote W, Kanla P, et al. Chromosomes and karyotype analysis of a liver fluke, Opisthorchis viverrini, by scanning electron microscopy.Parasitol Int · 2012 · 61(3):504-7doi:10.1016/j.parint.2012.03.008 · PMID 22504416

    Independent confirmation of 2n=12, with chromosomes prepared from the testes of adult worms because that is where dividing cells are. The constraint on karyotyping, illustrated rather than asserted.

21. 21
    Mutafova T, Gergova S. Cytological studies on three hymenolepidid species.J Helminthol · 1994 · 68(4):323-5doi:10.1017/s0022149x00001577 · PMID 7706680

    2n=12 for the dwarf tapeworm and two relatives. The record uses the genus Vampirolepis, which is why a search on Hymenolepis alone misses it.

22. 22
    Wendt GR, Shiroor DA, Adler CE, Collins JJ. Convergent evolution of a genotoxic stress response in a parasite-specific p53 homolog.Proc Natl Acad Sci U S A · 2022 · 119(37):e2205201119doi:10.1073/pnas.2205201119 · PMID 36067283

    Two p53 homologs in a parasitic flatworm: a true ortholog of planarian p53 and human TP53, and a parasite-specific paralog needed for the genotoxic stress response. Animal RNA-interference work; the PubMed record has had gene symbols stripped, so none are quoted from it.

23. 23
    Wendt GR, Collins JJ. Unusual inheritance of a functional cki homolog in the human pathogen Schistosoma mansoni.Sci Adv · 2025 · 11(50):eaea4905doi:10.1126/sciadv.aea4905 · PMID 41370391

    Knockdown of a schistosome cyclin-dependent kinase inhibitor homolog caused hyperproliferation, and double knockdown with the p53 ortholog produced tumor-like growths; the horizontal-transfer reading comes from the gene's absence in free-living ancestors. The record's title and abstract have had gene and species names stripped; the preprint [115] and a 2026 commentary [116] identify the gene as cki and the p53 as p53-1, in Schistosoma mansoni, and the full text names the gene Smp_199050.

24. 24
    Osman A, Niles EG, LoVerde PT. Characterization of the Ras homologue of Schistosoma mansoni.Mol Biochem Parasitol · 1999 · 100(1):27-41doi:10.1016/s0166-6851(99)00029-8 · PMID 10376991

    81% identity and 92% similarity to K-Ras, single copy, 23 kDa, farnesylated, with processing blocked by a farnesyl transferase inhibitor. The figures are protein-level; the nucleotide identity that governs primer annealing is not reported.

25. 25
    Spiliotis M, Tappe D, Brückner S, et al. Molecular cloning and characterization of Ras- and Raf-homologues from the fox-tapeworm Echinococcus multilocularis.Mol Biochem Parasitol · 2005 · 139(2):225-37doi:10.1016/j.molbiopara.2004.11.013 · PMID 15664657

    EmRas at 79% identical residues to human Ras, complementing the yeast gene in a growth assay, plus EmRaf sharing four conserved exon-intron boundaries with the human gene for Raf-1. A Raf-family kinase, not a BRAF ortholog.

26. 26
    Shoemaker CB, Ramachandran H, Landa A, et al. Alternative splicing of the Schistosoma mansoni gene encoding a homologue of epidermal growth factor receptor.Mol Biochem Parasitol · 1992 · 53(1-2):17-32doi:10.1016/0166-6851(92)90003-3 · PMID 1501637

    A 200-kDa EGFR homolog with at least three alternatively spliced variants, recovered by homology to the erbB tyrosine kinase domain. Clinical EGFR mutation tests also read the kinase domain [122], so this paper does not show divergence at the tested positions.

27. 27
    Beckmann S, Hahnel S, Cailliau K, et al. Characterization of the Src/Abl hybrid kinase SmTK6 of Schistosoma mansoni.J Biol Chem · 2011 · 286(49):42325-42336doi:10.1074/jbc.M110.210336 · PMID 22013071

    A schistosome kinase with a hybrid character and an inhibition profile intermediate between Src and Abl inhibitors, assayed in frog oocytes. Biochemistry in a heterologous system.

28. 28
    Loos JA, Caparros PA, Nicolao MC, et al. Identification and pharmacological induction of autophagy in the larval stages of Echinococcus granulosus: an active catabolic process in calcareous corpuscles.Int J Parasitol · 2014 · 44(7):415-27doi:10.1016/j.ijpara.2014.02.007 · PMID 24703869

    TORC1 expression and rapamycin sensitivity in tapeworm larvae, with autophagy genes identified in the genome. Evidence that the pathway exists in the parasite; not a treatment study.

29. 29
    Smout MJ, Laha T, Mulvenna J, et al. A granulin-like growth factor secreted by the carcinogenic liver fluke, Opisthorchis viverrini, promotes proliferation of host cells.PLoS Pathog · 2009 · 5(10):e1000611doi:10.1371/journal.ppat.1000611 · PMID 19816559

    The parasite's secreted granulin stimulated murine fibroblasts at nanomolar concentrations, with the effect blocked by a MAPK kinase inhibitor and by antibodies, and is the main growth factor in the fluke's secretions. In vitro and in infected hamsters; a protein crossing into the host, not a gene appearing in a test. The discoverers are now developing Ov-GRN-1 peptides as wound-healing agents [118].

30. 30
    Ayyadevara S, Thaden JJ, Shmookler Reis RJ. Discrimination of primer 3'-nucleotide mismatch by taq DNA polymerase during polymerase chain reaction.Anal Biochem · 2000 · 284(1):11-8doi:10.1006/abio.2000.4635 · PMID 10933850

    40- to 100-fold discrimination against a single mismatch at a primer's 3'-terminal T, G or C, and 8- to 20-fold one base in. The quantitative basis for the inference about human primers on parasite template; measured in a controlled reaction, not in a clinical assay.

31. 31
    Akkari Y, Baughn LB, Kim A, et al. Section E6.1-6.6 of the American College of Medical Genetics and Genomics (ACMG) Technical Laboratory Standards: Cytogenomic studies of acquired chromosomal abnormalities in neoplastic blood, bone marrow, and lymph nodes.Genet Med · 2024 · 26(4):101054doi:10.1016/j.gim.2023.101054 · PMID 38349293

    The current American laboratory standard for cancer cytogenetics in blood, marrow and lymph node. A standards document; it does not report accuracy data.

32. 32
    Döhner H, Wei AH, Appelbaum FR, et al. Diagnosis and management of AML in adults: 2022 recommendations from an international expert panel on behalf of the ELN.Blood · 2022 · 140(12):1345-1377doi:10.1182/blood.2022016867 · PMID 35797463

    Expert-panel guideline building a revised genetic risk classification for acute myeloid leukemia on cytogenetics together with molecular markers. Individual author disclosures are not in the PubMed record and should not be assumed absent.

33. 33
    Sosinsky A, Ambrose J, Cross W, et al. Insights for precision oncology from the integration of genomic and clinical data of 13,880 tumors from the 100,000 Genomes Cancer Programme.Nat Med · 2024 · 30(1):279-289doi:10.1038/s41591-023-02682-0 · PMID 38200255

    Whole-genome sequencing of 13,880 solid tumors across 33 cancer types with linked outcome data: 94% of glioblastomas with small variants, 58% with copy-number aberrations, 13% of sarcomas with actionable structural variants, 40% homologous recombination deficiency in high-grade serous ovarian cancer. Observational, within one health system.

34. 34
    Sepulveda AR, Hamilton SR, Allegra CJ, et al. Molecular Biomarkers for the Evaluation of Colorectal Cancer: Guideline From the American Society for Clinical Pathology, College of American Pathologists, Association for Molecular Pathology, and the American Society of Clinical Oncology.J Clin Oncol · 2017 · 35(13):1453-1486doi:10.1200/JCO.2016.71.9807 · PMID 28165299

    Twenty-one statements from a search of over 4,000 articles, recommending mutational testing of epidermal growth factor receptor pathway genes. Two panel members list commercial molecular-diagnostics affiliations, Castle Biosciences and Biocept.

35. 35
    Cimino PJ, Zhao G, Wang D, Sehn JK, Lewis JS, Duncavage EJ. Detection of viral pathogens in high grade gliomas from unmapped next-generation sequencing data.Exp Mol Pathol · 2014 · 96(3):310-5doi:10.1016/j.yexmp.2014.03.010 · PMID 24704430

    Twenty-one consecutive formalin-fixed high-grade gliomas on a 151-gene clinical oncology panel; an average of 38,000 non-human reads per case (1.9%) were aligned to the NCBI nucleotide databases and classified by taxonomy; Epstein-Barr virus in 5 of 21 and Roseolovirus in 1 of 21, no cytomegalovirus, and EBV in situ hybridization negative in all four cases with material. The senior author of this paper is the first author of [18]. One laboratory, 21 cases.

36. 36
    Petrova-Drus K, Quesada AE, Bowman AS, et al. Quantitative Off-Target Detection of Epstein-Barr Virus-Derived DNA in Routine Molecular Profiling of Hematopoietic Neoplasms by Panel-Based Hybrid-Capture Next-Generation Sequencing.J Mol Diagn · 2022 · 24(1):69-78doi:10.1016/j.jmoldx.2021.10.009 · PMID 34801704

    5,234 samples from 3,636 patients on a 400-gene hybridization-capture panel: 100 samples (1.9%) in 93 patients carried 6 or more off-target EBV reads (range 6 to 32,325), 73% of them known EBV-associated conditions; Spearman r = 0.8 (95% CI 0.74 to 0.85) against EBV-encoded RNA in situ hybridization in 162 samples, with 10 or more reads the best threshold. An institutional validation; the panel was not designed to find the virus.

37. 37
    Miller DT, Adam MP, Aradhya S, et al. Consensus statement: chromosomal microarray is a first-tier clinical diagnostic test for individuals with developmental disabilities or congenital anomalies.Am J Hum Genet · 2010 · 86(5):749-64doi:10.1016/j.ajhg.2010.04.006 · PMID 20466091

    The 15-20% versus about 3% yield figures, from 33 studies and 21,698 patients, plus the statement that balanced rearrangements and low-level mosaicism are generally not detectable by arrays. Developmental disorders, not cancer.

38. 38
    Ledbetter DH, Martin CL. Cryptic telomere imbalance: a 15-year update.Am J Med Genet C Semin Med Genet · 2007 · 145C(4):327-34doi:10.1002/ajmg.c.30149 · PMID 17910073

    Source for cryptic telomeric imbalance in 3-6% of unexplained intellectual disability, for roughly 40% of those imbalances exceeding 5 megabases, and for the explicit conclusion that the G-banded karyotype's analytic sensitivity is much lower than previously thought.

39. 39
    Duncavage EJ, Schroeder MC, O'Laughlin M, et al. Genome Sequencing as an Alternative to Cytogenetic Analysis in Myeloid Cancers.N Engl J Med · 2021 · 384(10):924-935doi:10.1056/NEJMoa2024534 · PMID 33704937

    263 patients; whole-genome sequencing detected all 40 recurrent translocations and 91 copy-number alterations found by cytogenetics, added reportable events in 40 of 235, and changed risk category for 19 of 117 prospective cases. A single-center feasibility and accuracy study, not a randomized trial of outcomes.

40. 40
    Sripa B, Leungwattanawanit S, Nitta T, et al. Establishment and characterization of an opisthorchiasis-associated cholangiocarcinoma cell line (KKU-100).World J Gastroenterol · 2005 · 11(22):3392-7doi:10.3748/wjg.v11.i22.3392 · PMID 15948244

    A parasite-associated tumor that was karyotyped, but only after four months of culture produced dividing cells: modal chromosome number 78, and immunocytochemistry negative for p53. One patient.

41. 41
    Prior IA, Lewis PD, Mattos C. A comprehensive survey of Ras mutations in cancer.Cancer Res · 2012 · 72(10):2457-67doi:10.1158/0008-5472.CAN-11-2612 · PMID 22589270

    The Ras proteins promote oncogenesis when mutationally activated at codon 12, 13 or 61, with distinctive codon and substitution biases between isoforms. Cited for which codons the hotspot is, from curated large-scale tumor profiling.

42. 42
    Chakravarty D, Johnson A, Sklar J, et al. Somatic Genomic Testing in Patients With Metastatic or Advanced Cancer: ASCO Provisional Clinical Opinion.J Clin Oncol · 2022 · 40(11):1231-1258doi:10.1200/JCO.21.02767 · PMID 35175857

    The American counterpart to [29]: sequencing in a certified laboratory when an approved biomarker exists, multigene panels when more than one biomarker-linked therapy is approved. Consensus; individual disclosures not in the record.

43. 43
    Pinkel D, Straume T, Gray JW. Cytogenetic analysis using quantitative, high-sensitivity, fluorescence hybridization.Proc Natl Acad Sci U S A · 1986 · 83(9):2934-8doi:10.1073/pnas.83.9.2934 · PMID 3458254

    The founding FISH paper: staining in metaphase spreads and interphase nuclei, and male cells forty times brighter than female with a 0.8-kilobase Y probe. A methods demonstration in cell lines and amniocytes.

44. 44
    Muehlenbachs A, Bhatnagar J, Agudelo CA, et al. Malignant Transformation of Hymenolepis nana in a Human Host.N Engl J Med · 2015 · 373(19):1845-52doi:10.1056/NEJMoa1505892 · PMID 26535513

    The single strongest datum for the season's thesis, and the one that shows which assay was needed: a polymerase chain reaction targeting eukaryotes, with immunohistochemistry and probe hybridization localizing worm material in the cells. One patient, described by the authors as a novel disease mechanism.

45. 45
    Santamaría-Fríes M, Fajardo LF, Sogin ML, et al. Lethal infection by a previously unrecognised metazoan parasite.Lancet · 1996 · 347(9018):1797-801doi:10.1016/s0140-6736(96)91618-9 · PMID 8667924

    An abdominal mass suggestive of a neoplasm in a man with HIV, identified as an uncharacterized cestode by amplifying 357 base pairs of ribosomal DNA from the tissue. The precedent for broad-range amplification succeeding where morphology failed; one patient.

46. 46
    Yunis JJ. High resolution of human chromosomes.Science · 1976 · 191(4233):1268-70doi:10.1126/science.1257746 · PMID 1257746

    High-resolution human chromosome preparations. No abstract in the record; cited for the advance, not for any number.

47. 47
    Nowell PC, Hungerford DA. Chromosome studies on normal and leukemic human leukocytes.J Natl Cancer Inst · 1960 · 25:85-109PMID 14427847

    The primary 1960 report of the chromosome studies behind the Philadelphia chromosome. No abstract or DOI in the record; cited for the date and authorship, with the detail taken from [9].

48. 48
    Rowley JD. A new consistent chromosomal abnormality in chronic myelogenous leukaemia identified by quinacrine fluorescence and Giemsa staining.Nature · 1973 · 243(5405):290-3doi:10.1038/243290a0 · PMID 4126434

    The primary report identifying the abnormality as a rearrangement. The record carries no abstract, so the detail of the 9;22 exchange is taken from the reviews cited alongside it.

49. 49
    Tefferi A. The history of myeloproliferative disorders: before and after Dameshek.Leukemia · 2008 · 22(1):3-13doi:10.1038/sj.leu.2404946 · PMID 17882283

    Review supplying the dates: the Philadelphia chromosome in 1960, its decipherment as a 9;22 reciprocal translocation in 1972, and imatinib in 1996. A narrative review, not primary data.

50. 50
    Kallioniemi A, Kallioniemi OP, Sudar D, et al. Comparative genomic hybridization for molecular cytogenetic analysis of solid tumors.Science · 1992 · 258(5083):818-21doi:10.1126/science.1359641 · PMID 1359641

    Genome-wide copy number from two differentially labeled DNAs; 16 amplified regions in bladder tumors and cell lines. The ancestor of the clinical microarray, not itself a clinical test.

51. 51
    Ng SB, Turner EH, Robertson PD, et al. Targeted capture and massively parallel sequencing of 12 human exomes.Nature · 2009 · 461(7261):272-6doi:10.1038/nature08250 · PMID 19684571

    The demonstration that all protein-coding sequence could be captured and read, across twelve people and more than 300 megabases of coding sequence. A proof of concept in Mendelian disease, not oncology.

52. 52
    Ley TJ, Mardis ER, Ding L, et al. DNA sequencing of a cytogenetically normal acute myeloid leukaemia genome.Nature · 2008 · 456(7218):66-72doi:10.1038/nature07485 · PMID 18987736

    The first whole-genome sequence of a human cancer: one AML genome against the patient's skin; ten genes with acquired mutations, eight new. A single patient, in a leukemia the karyotype had called normal.

53. 53
    Neveling K, Mantere T, Vermeulen S, et al. Next-generation cytogenetics: Comprehensive assessment of 52 hematological malignancy genomes by optical genome mapping.Am J Hum Genet · 2021 · 108(8):1423-1435doi:10.1016/j.ajhg.2021.06.001 · PMID 34237281

    Detected all clinically reported aberrations in 36 simple cases and was largely concordant in 16 complex ones, with 100% sensitivity and positive predictive value above 80% in technical comparison; the authors present it as a potential replacement for the karyotype-FISH-microarray combination. A single commercial platform is assessed.

54. 54
    Levy B, Baughn LB, Akkari Y, et al. Optical genome mapping in acute myeloid leukemia: a multicenter evaluation.Blood Adv · 2023 · 7(7):1297-1307doi:10.1182/bloodadvances.2022007583 · PMID 36417763

    100 AML cases previously characterized by karyotype alone or with FISH or microarray: all clinically relevant structural and copy-number variants recovered when the clone exceeded 5% allelic fraction; clinically relevant information missed by routine methods in 13%; three normal karyotypes carried cryptic fusions; management would have changed in 4%. Two authors at a commercial laboratory (PathGroup).

55. 55
    Levy B, Kanagal-Shamanna R, Sahajpal NS, et al. A framework for the clinical implementation of optical genome mapping in hematologic malignancies.Am J Hematol · 2024 · 99(4):642-661doi:10.1002/ajh.27175 · PMID 38164980

    Consensus framework on validation, quality control and variant interpretation from the International Consortium for OGM in Hematologic Malignancies; one author is employed by Bionano, the platform's manufacturer.

56. 56
    Kanagal-Shamanna R, Puiggros A, Granada I, et al. Integration of Optical Genome Mapping in the Cytogenomic and Molecular Work-Up of Hematological Malignancies: Expert Recommendations From the International Consortium for Optical Genome Mapping.Am J Hematol · 2025 · 100(6):1029-1048doi:10.1002/ajh.27688 · PMID 40304265

    Expert recommendations on clinical indications, testing algorithms and interpretation, framed as integrating OGM as a standard-of-care cytogenetic assay; a consensus document from the method's adopters, not an outcomes study.

57. 57
    Feng H, Shuda M, Chang Y, Moore PS. Clonal integration of a polyomavirus in human Merkel cell carcinoma.Science · 2008 · 319(5866):1096-100doi:10.1126/science.1152586 · PMID 18202256

    Non-human sequence found in tumor data and shown to matter: a 5,387-base-pair virus in 8 of 10 tumors against 5 of 59 control tissues, clonally integrated in 6 of 8. Small numbers, and the causal claim is hedged by the authors.

58. 58
    Conn DB. Malignant Transformation of Hymenolepis nana in a Human Host.N Engl J Med · 2016 · 374(13):1293doi:10.1056/NEJMc1600490 · PMID 27028926

    Correspondence on the 2015 case, answered by the authors in [92]. Cited only to show that what followed the report was discussion, not replication; the record has no abstract.

59. 59
    Muehlenbachs A, Mathison BA, Olson PD. Malignant Transformation of Hymenolepis nana in a Human Host [reply].N Engl J Med · 2016 · 374(13):1293-4doi:10.1056/NEJMc1600490 · PMID 27028925

    The original authors' reply to the correspondence on the 2015 case. The record carries no abstract; cited to show that what followed the report was discussion, not replication.

60. 60
    Poore GD, Kopylova E, Zhu Q, et al. Microbiome analyses of blood and tissues suggest cancer diagnostic approach.Nature · 2020 · 579(7800):567-574doi:10.1038/s41586-020-2095-1 · PMID 32214244

    RETRACTED. 18,116 archived tissue and blood samples across 33 cancer types, with up to 92.3% of sequence data discarded in decontamination and a claimed plasma test separating 69 cancer-free people from 100 samples from patients with prostate cancer, lung cancer or melanoma. Cite only as a retracted paper; one co-author lists a commercial genomics affiliation and the paper proposed a commercializable diagnostic. The authors' defense is [99], and the re-analysis has since been extended to all 5,734 Atlas genomes [101].

61. 61
    Aganezov S, Yan SM, Soto DC, et al. A complete reference genome improves analysis of human genetic variation.Science · 2022 · 376(6588):eabl3533doi:10.1126/science.abl3533 · PMID 35357935

    T2T-CHM13 against 3,202 short-read and 17 long-read samples: improved mapping and variant calling, hundreds of thousands of new variants per sample, tens of thousands of spurious variants removed, false positives in 269 medically relevant genes reduced up to twelvefold. Co-authors at Bionano Genomics and DNAnexus.

62. 62
    Poore GD, Kopylova E, Zhu Q, et al. Retraction Note: Microbiome analyses of blood and tissues suggest cancer diagnostic approach.Nature · 2024 · 631(8021):694doi:10.1038/s41586-024-07656-x · PMID 38926587

    The formal retraction, four years after publication. The record carries no abstract, so the stated reason for retraction is not characterized here.

63. 63
    Nejman D, Livyatan I, Fuks G, et al. The human tumor microbiome is composed of tumor type-specific intracellular bacteria.Science · 2020 · 368(6494):973-980doi:10.1126/science.aay9189 · PMID 32467386

    1,526 tumors and adjacent normal tissue across seven cancer types, reporting distinct compositions and mostly intracellular bacteria, with the low biomass named as the central difficulty. Correlational.

64. 64
    Ge Y, Lu J, Puiu D, Revsine M, Salzberg SL. Comprehensive analysis of microbial content in whole-genome sequencing samples from The Cancer Genome Atlas project.Sci Transl Med · 2025 · 17(814):eads6335doi:10.1126/scitranslmed.ads6335 · PMID 40901923

    All 5,734 TCGA whole-genome datasets across 25 cancer types, re-classified with updated methods and databases and compared with two major earlier studies of bacteria, viruses and fungi in cancer; microbial presence far smaller than reported, many species possibly absent; per-sample read counts released. From the same group as [46]; parasites were not among the taxa counted.

65. 65
    Sepich-Poore GD, McDonald D, Kopylova E, et al. Robustness of cancer microbiome signals over a broad range of methodological variation.Oncogene · 2024 · 43(15):1127-1148doi:10.1038/s41388-024-02974-w · PMID 38396294

    The original authors' defense of the retracted 2020 paper: batch-correction equivalence, about 1% database contamination by their count, a new RefSeq-cleaning method, and persistence of cancer type-specific signals under the critics' pipeline. First author and several others list Micronoma, a commercial microbial-diagnostics company; one lists Clarity Genomics. The journal that published it did not retract it.

66. 66
    Zehir A, Benayed R, Shah RH, et al. Mutational landscape of metastatic cancer revealed from prospective clinical sequencing of 10,000 patients.Nat Med · 2017 · 23(6):703-713doi:10.1038/nm.4333 · PMID 28481359

    More than 10,000 prospectively sequenced patients, 11% enrolled on genomically matched trials. Shows the scale of panel testing; says nothing about non-human sequence.

67. 67
    Zehir A, Benayed R, Shah RH, et al. Erratum: Mutational landscape of metastatic cancer revealed from prospective clinical sequencing of 10,000 patients.Nat Med · 2017 · 23(8):1004doi:10.1038/nm0817-1004c · PMID 28777785

    Published erratum to [26], cited with it on the principle that a source is quoted with its amendment. The record carries no abstract, so its content is not characterized here.

68. 68
    Wolff AC, Hammond MEH, Allison KH, et al. Human Epidermal Growth Factor Receptor 2 Testing in Breast Cancer: American Society of Clinical Oncology/College of American Pathologists Clinical Practice Guideline Focused Update.J Clin Oncol · 2018 · 36(20):2105-2122doi:10.1200/JCO.2018.77.8738 · PMID 29846122

    Cited for the statement that roughly 5% of cases fall into dual-probe in situ hybridization groups that require concomitant immunohistochemistry to call. A consensus guideline.

69. 69
    Pascual J, Attard G, Bidard FC, et al. ESMO recommendations on the use of circulating tumour DNA assays for patients with cancer: a report from the ESMO Precision Medicine Working Group.Ann Oncol · 2022 · 33(8):750-768doi:10.1016/j.annonc.2022.05.520 · PMID 35809752

    Sets out what plasma DNA testing may and may not be used for, including poor detection of fusions and copy-number change and the advice to reflex to tissue after a non-informative result. Consensus, not an accuracy study.

70. 70
    Lockwood CM, Messersmith HJ, Kim AS, et al. Circulating Tumor DNA Testing in Solid Tumors and Lymphoma: ASCO Guideline.JCO Oncol Pract · 2026 · 22(9):1468-1489doi:10.1200/OP-26-00311 · PMID 42314080

    Systematic review (January 2017 to February 2025) by a multidisciplinary panel with patient representation; use when tissue testing is challenging, unsafe or too slow or when a regulatory indication allows; tissue confirmation after negative, inconclusive or inconsistent results; ctDNA fraction not a surrogate. Individual disclosures are not in the record.

71. 71
    Razavi P, Li BT, Brown DN, et al. High-intensity sequencing reveals the sources of plasma circulating cell-free DNA variants.Nat Med · 2019 · 25(12):1928-1937doi:10.1038/s41591-019-0652-7 · PMID 31768066

    508 genes at more than 60,000-fold raw depth in 124 patients and 47 controls: 53.2% of plasma mutations in patients and 81.6% in controls were consistent with clonal hematopoiesis rather than tumor. Several authors were employed by a commercial sequencing company.

72. 72
    Mikhail FM, Heerema NA, Rao KW, et al. Section E6.1-6.4 of the ACMG technical standards and guidelines: chromosome studies of neoplastic blood and bone marrow-acquired chromosomal abnormalities.Genet Med · 2016 · 18(6):635-42doi:10.1038/gim.2016.50 · PMID 27124785

    The 2016 edition superseded by the 2024 standards, cited to show the line of succession. One author is affiliated with a commercial reference laboratory.

73. 73
    Mosele MF, Westphalen CB, Stenzinger A, et al. Corrigendum to "Recommendations for the use of next-generation sequencing (NGS) for patients with advanced cancer in 2024: a report from the ESMO Precision Medicine Working Group".Ann Oncol · 2025 · 36(4):472doi:10.1016/j.annonc.2024.11.010 · PMID 39984357

    The published correction to the 2024 recommendation, cited so the guideline is quoted with its own amendment. The record carries no abstract, so the content of the correction is not described here.

74. 74
    Schneider VA, Graves-Lindsay T, Howe K, et al. Evaluation of GRCh38 and de novo haploid genome assemblies demonstrates the enduring quality of the reference assembly.Genome Res · 2017 · 27(5):849-864doi:10.1101/gr.213611.116 · PMID 28396521

    Documents GRCh38 as a construction under repair: roughly a thousand issues resolved, the first sequence representation of centromeres, and altered read alignments even in unchanged regions.

75. 75
    Breitwieser FP, Pertea M, Zimin AV, Salzberg SL. Human contamination in bacterial genomes has created thousands of spurious proteins.Genome Res · 2019 · 29(6):954-960doi:10.1101/gr.245373.118 · PMID 31064768

    2,250 bacterial and archaeal assemblies contaminated with human sequence, generating 3,437 spurious protein entries. Explains how human reads come to match bacteria; it is a database audit, not a clinical study.

76. 76
    Steinegger M, Salzberg SL. Terminating contamination: large-scale search identifies more than 2,000,000 contaminated entries in GenBank.Genome Biol · 2020 · 21(1):115doi:10.1186/s13059-020-02023-1 · PMID 32398145

    2,161,746 contaminated sequences in RefSeq and 114,035 in GenBank. The scale of mislabeled reference sequence that any taxonomic search inherits.

77. 77
    Kostic AD, Ojesina AI, Pedamallu CS, et al. PathSeq: software to identify or discover microbes by deep sequencing of human tissue.Nat Biotechnol · 2011 · 29(5):393-6doi:10.1038/nbt.1868 · PMID 21552235

    The computational subtraction approach for finding microbes in human tissue sequence. A short methods letter with no abstract in the record; cited for the architecture, host subtraction first.

78. 78
    Frolova M, Maguire B, Duessmann H, et al. HumanFilt: a multi-reference host depletion pipeline improves Fusobacterium detection accuracy in tumor WGS data sets.mSystems · 2026 · 11(8):e0055026doi:10.1128/msystems.00550-26 · PMID 42446235

    After k-mer, vector and duplicate filtering, three aligner-and-reference pairs collectively remove more than 99.9% of human reads, which matters because non-human reads are a small minority in tumor data; in simulations it kept more than 99.7% of true bacterial signal, at the cost of removing more ambiguous PathSeq-classified reads than two comparison tools. Ten mucinous rectal adenocarcinomas; a methods paper.

79. 79
    Castellarin M, Warren RL, Freeman JD, et al. Fusobacterium nucleatum infection is prevalent in human colorectal carcinoma.Genome Res · 2012 · 22(2):299-306doi:10.1101/gr.126516.111 · PMID 22009989

    RNA sequencing with host subtraction, then confirmation by quantitative PCR in 99 subjects. Association with tumor tissue and with lymph node metastasis; not causation.

80. 80
    Bullman S, Pedamallu CS, Sicinska E, et al. Analysis of Fusobacterium persistence and antibiotic response in colorectal cancer.Science · 2017 · 358(6369):1443-1448doi:10.1126/science.aal5240 · PMID 29170280

    The bacterium persists in distal metastases and in mouse xenografts, and an antibiotic reduced bacterial load and tumor growth. The treatment experiment is in mice, not patients.

81. 81
    Zapatka M, Borozan I, Brewer DS, et al. The landscape of viral associations in human cancers.Nat Genet · 2020 · 52(3):320-330doi:10.1038/s41588-019-0558-9 · PMID 32025001

    A consensus of three independent pipelines across 2,658 cancer genomes found viruses in 382 genome and 68 transcriptome datasets, overwhelmingly known tumor viruses. Shows what a careful search of tumor sequence does and does not turn up.

82. 82
    Zapatka M, Borozan I, Brewer DS, et al. Author Correction: The landscape of viral associations in human cancers.Nat Genet · 2023 · 55(6):1077doi:10.1038/s41588-023-01316-y · PMID 36944734

    Published author correction to [42], cited with it. The record carries no abstract, so its content is not characterized here.

83. 83
    Salter SJ, Cox MJ, Turek EM, et al. Reagent and laboratory contamination can critically impact sequence-based microbiome analyses.BMC Biol · 2014 · 12:87doi:10.1186/s12915-014-0087-z · PMID 25387460

    Contaminating DNA is ubiquitous in extraction kits, varies between kits and batches, and dominates low-biomass samples; negative controls are strongly advised. The mundane explanation that any parasite-DNA claim must clear first.

84. 84
    Costello M, Fleharty M, Abreu J, et al. Characterization and remediation of sample index swaps by non-redundant dual indexing on massively parallel sequencing platforms.BMC Genomics · 2018 · 19(1):332doi:10.1186/s12864-018-4703-0 · PMID 29739332

    Index swapping at 0.2% to 6% across named 2018 instruments, with non-redundant dual indexing as the fix. The rates are instrument- and era-specific and should not be quoted as today's clinical figure.

85. 85
    de Goffau MC, Lager S, Sovio U, et al. Human placenta has no microbiome but can contain potential pathogens.Nature · 2019 · 572(7769):329-334doi:10.1038/s41586-019-1451-5 · PMID 31367035

    A worked example of a low-biomass microbiome that dissolved on careful analysis, with almost all signal traceable to labor or reagent contamination, except group B Streptococcus in about 5% of pre-labor samples.

86. 86
    Narunsky-Haziza L, Sepich-Poore GD, Livyatan I, et al. Pan-cancer analyses reveal cancer-type-specific fungal ecologies and bacteriome interactions.Cell · 2022 · 185(20):3789-3806.e17doi:10.1016/j.cell.2022.09.005 · PMID 36179670

    Fungal DNA and cells at low abundance across 17,401 samples and 35 cancer types, with histological staining of tissue arrays; prognostic and diagnostic claims for tissue and plasma mycobiomes. Second author and others list Micronoma; shares senior authorship with the retracted paper [44]. Not retracted.

87. 87
    Dohlman AB, Klug J, Mesko M, et al. A pan-cancer mycobiome analysis reveals fungal involvement in gastrointestinal and lung tumors.Cell · 2022 · 185(20):3807-3822.e12doi:10.1016/j.cell.2022.09.015 · PMID 36179671

    Tumor-associated fungi at very low abundance; Candida enriched across gastrointestinal sites, predictive of metastasis in colon cancer and of decreased survival, confirmed by independent ITS sequencing and culture in a separate cohort. Association, not causation; the authors call fungal DNA a candidate biomarker.

88. 88
    Sepich-Poore GD, McDonald D, Kopylova E, et al. Correction: Robustness of cancer microbiome signals over a broad range of methodological variation.Oncogene · 2024 · 43(20):1579doi:10.1038/s41388-024-03018-z · PMID 38580705

    Published erratum to the rebuttal; the record carries no abstract, so its content is not characterized here.

89. 89
    Salzberg SL, Chia M, Tay A, et al. Setting higher standards for reports of microbial species in human cancers.Nat Cancer · 2026 · 7(3):414-423doi:10.1038/s43018-026-01121-6 · PMID 41714823

    A standards and quality-control review for tumor-microbiome reports, naming sampling, storage, handling, processing and contaminated reference genomes as sources of false microbial signal, by analogy with ancient-DNA work. A review by one side of the dispute.

90. 90
    Leiendecker L, Neumann T, Jung PS, et al. Human Papillomavirus 42 Drives Digital Papillary Adenocarcinoma and Elicits a Germ Cell-like Program Conserved in HPV-Positive Cancers.Cancer Discov · 2023 · 13(1):70-84doi:10.1158/2159-8290.CD-22-0489 · PMID 36213965

    Off-target reads from commonly available next-generation sequencing data across 19 skin tumor types; HPV42 in 45 of 47 digital papillary adenocarcinomas. A discovery of a new tumor virus from reads a human-targeted assay had not asked for; one author lists a company affiliation (Quantro Therapeutics).

91. 91
    Lawrence L, Wang A, Charville G, et al. Identification and confirmation via in situ hybridization of Merkel cell polyomavirus in rare cases of posttransplant cutaneous T-cell lymphoma.J Cutan Pathol · 2023 · 50(9):835-844doi:10.1111/cup.14486 · PMID 37394808

    Off-target reads from a targeted hematolymphoid panel aligned to a viral genome database; Merkel cell polyomavirus DNA in four post-transplant folliculotropic mycosis fungoides and one peripheral T-cell lymphoma, with viral RNA localized by in situ hybridization. Five cases; two authors at Guardant Health, a commercial sequencing company.

92. 92
    Koonin EV. Orthologs, paralogs, and evolutionary genomics.Annu Rev Genet · 2005 · 39:309-38doi:10.1146/annurev.genet.39.073003.114725 · PMID 16285863

    The definitions the ortholog argument depends on, and the reason the distinction matters for inferring function. A review, not evidence about parasites.

93. 93
    Belyi VA, Ak P, Markert E, et al. The origins and evolution of the p53 family of genes.Cold Spring Harb Perspect Biol · 2010 · 2(6):a001198doi:10.1101/cshperspect.a001198 · PMID 20516129

    The family ancestor is first detected in sea anemones, is present in almost all invertebrates, and has preserved structure and function for over a billion years. Also the point at which the homology argument must stop: a billion years is a billion years of sequence change.

94. 94
    Wendt G, Collins JJ. Horizontal gene transfer of a functional cki homolog in the human pathogen Schistosoma mansoni [preprint].bioRxiv · 2024 · 2024.05.27.596073doi:10.1101/2024.05.27.596073 · PMID 38853947

    The May 2024 preprint of [67]; cki knockdown gives hyperproliferation, double knockdown with p53 gives tumor-like growths. Not peer reviewed; the PubMed title has the gene and species names stripped, restored here from [116].

95. 95
    Brindley PJ. Tumor-suppressor pathways in Schistosoma mansoni support a novel hypothesis on neodermatan flatworm origins.Trends Parasitol · 2026 · 42(4):212-214doi:10.1016/j.pt.2026.02.003 · PMID 41820076

    Commentary on [67]: names cki and p53-1, summarizes the tegument hyperproliferation and tumor-like growths, and reads the cki distribution as horizontal gene transfer aiding the evolution of parasitism. One author's interpretation.

96. 96
    Lynch TJ, Bell DW, Sordella R, et al. Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib.N Engl J Med · 2004 · 350(21):2129-39doi:10.1056/NEJMoa040938 · PMID 15118073

    The activating EGFR mutations that clinical testing looks for are small in-frame deletions and substitutions in the tyrosine kinase domain, clustered around the ATP-binding pocket; found in 8 of 9 gefitinib responders and none of 7 non-responders. Cited for where in the gene the tested positions lie, which is the same conserved domain through which the fluke homolog was cloned [70].

97. 97
    Beckmann S, Grevelding CG. Imatinib has a fatal impact on morphology, pairing stability and survival of adult Schistosoma mansoni in vitro.Int J Parasitol · 2010 · 40(5):521-6doi:10.1016/j.ijpara.2010.01.007 · PMID 20149792

    The drug that came out of the Philadelphia chromosome story damages and kills adult worms in culture. In vitro only, with no animal or human treatment data.

98. 98
    Muhedier M, Li J, Liu H, et al. Tacrolimus, a rapamycin target protein inhibitor, exerts anti-cystic echinococcosis effects both in vitro and in vivo.Acta Trop · 2020 · 212:105708doi:10.1016/j.actatropica.2020.105708 · PMID 32956634

    Oral dosing at 4 mg/kg per day reduced cyst weight and number in mice, with ultrastructural damage. Mouse model of one parasite; no human data, and the drug is an immunosuppressant.

99. 99
    Papatpremsiri A, Smout MJ, Loukas A, et al. Suppression of Ov-grn-1 encoding granulin of Opisthorchis viverrini inhibits proliferation of biliary epithelial cells.Exp Parasitol · 2015 · 148:17-23doi:10.1016/j.exppara.2014.11.004 · PMID 25450776

    RNA interference suppressed expression by 95% at day 3 and about 100% by day 7, and co-culture with suppressed flukes cut hyperproliferation by 25% in normal cholangiocytes and 92% in a cancer line. Cell lines, with no animal or human outcome.

100. 100
     Ajmera S, Puckelwaldt O, Stroehlein AJ, Haeberlein S. Curation of the Fasciola hepatica kinome as a resource for drug target discovery.BMC Genomics · 2026 · 27(1):98doi:10.1186/s12864-025-12513-w · PMID 41545833

     245 protein kinases, 2.14% of the proteome; 25% of 110 pathway-annotated kinases on cancer pathways by orthology to human; vandetanib and ruboxistaurin lethal to immature flukes in vitro at 50 µM, ruboxistaurin reducing adult motility. A curation and in vitro screen.

101. 101
     McCusker P, Clarke NG, Armstrong R, et al. Polo-like kinase 1 regulates growth in juvenile Fasciola hepatica.PLoS Pathog · 2025 · 21(12):e1013406doi:10.1371/journal.ppat.1013406 · PMID 41329781

     271 putative kinases, about 1.6% of protein-coding genes; RNA interference against fhplk1 reduced growth and cell proliferation, phenocopied by the PLK inhibitor BI 2536; 946 genes down-regulated. Three authors at Boehringer Ingelheim Animal Health.

102. 102
     Ward P, Equinet L, Packer J, Doerig C. Protein kinases of the human malaria parasite Plasmodium falciparum: the kinome of a divergent eukaryote.BMC Genomics · 2004 · 5:79doi:10.1186/1471-2164-5-79 · PMID 15479470

     65 protein kinases, none clustering in the tyrosine kinase or STE groups, plus a 20-member family restricted to Apicomplexa. The hardest limit on the homology argument: a protozoan with no tyrosine-kinase-group kinases has no EGFR or KIT for a panel to find.

103. 103
     Arunsan P, Ittiprasert W, Smout MJ, et al. Programmed knockout mutation of liver fluke granulin attenuates virulence of infection-induced hepatobiliary morbidity.Elife · 2019 · 8:e41463doi:10.7554/eLife.41463 · PMID 30644359

     Gene-edited flukes still colonized hamster bile ducts but caused less biliary hyperplasia and fibrosis. The strongest causal link from a parasite gene to pre-malignant host pathology, and it is in hamsters.

104. 104
     Blauwkamp TA, Thair S, Rosen MJ, et al. Analytical and clinical validation of a microbial cell-free DNA sequencing test for infectious disease.Nat Microbiol · 2019 · 4(4):663-674doi:10.1038/s41564-018-0349-6 · PMID 30742071

     A plasma sequencing test covering 1,250 organisms including eukaryotic parasites, with 93.7% agreement with blood culture in 350 patients. Almost all authors are employees of the company selling the test.

105. 105
     Zhou C, Li C, Deng Z, et al. Rapid diagnosis of alveolar echinococcosis from lung puncture sample using metagenomic next-generation sequencing: a case report.BMC Infect Dis · 2024 · 24(1):683doi:10.1186/s12879-024-09553-0 · PMID 38982338

     6,156 sequence reads matching Echinococcus multilocularis in a lung puncture sample, after pathology, sputum and stool had all failed. One patient; evidence of capability, not of accuracy.

106. 106
     Guo Y, Zhang J, Chai R, et al. Cerebral schistosomiasis in a 3-year-old girl due to Schistosoma japonicum: a case report.Front Immunol · 2024 · 15:1502627doi:10.3389/fimmu.2024.1502627 · PMID 39697332

     Brain lesions taken for an abscess, diagnosed by histology plus metagenomic sequencing of surgically obtained tissue. One patient, and the species name is stripped from the PubMed title.

107. 107
     Xie Y, Xu Y, Ma C, et al. Cerebral sparganosis caused by spirometra sp.: A case report.Diagn Microbiol Infect Dis · 2026 · 115(1):117311doi:10.1016/j.diagmicrobio.2026.117311 · PMID 41691882

     Seven years of migratory intracranial masses, resolved when metagenomic sequencing of cerebrospinal fluid named the parasite and tissue sequencing confirmed it. One patient; the kind of case that shows what is missed when nobody asks.

108. 108
     Benoit P, Brazer N, de Lorenzi-Tognon M, et al. Seven-year performance of a clinical metagenomic next-generation sequencing test for diagnosis of central nervous system infections.Nat Med · 2024 · 30(12):3522-3533doi:10.1038/s41591-024-03275-1 · PMID 39533109

     4,828 cerebrospinal-fluid samples, June 2016 to April 2023; 797 organisms in 697 samples, 23 of them parasites; sensitivity 63.1%, specificity 99.6%, accuracy 92.9% in 1,164 samples with clinical metadata; 21.8% of infectious diagnoses made by metagenomics alone. Five authors at Delve Bio, which commercializes the test.

109. 109
     Koonrungsesomboon N, Wadagni AC, Mbanefo EC. Molecular markers and Schistosoma-associated bladder carcinoma: A systematic review and meta-analysis.Cancer Epidemiol · 2015 · 39(4):487-96doi:10.1016/j.canep.2015.06.004 · PMID 26162479

     63 articles and 41 host molecular factors; elevated p53 and telomerase measured as host markers by design, against other schistosomiasis patients (worm p53 was not tested). Some intervals are extremely wide, so the narrower estimates are the quotable ones; association, not causation.

110. 110
     Bramwell G, DeGregori J, Thomas F, Ujvari B. Transmissible cancers, the genomes that do not melt down.Evolution · 2024 · 78(7):1205-1211doi:10.1093/evolut/qpae063 · PMID 38656785

     Eleven known transmissible cancer lineages, in dogs, Tasmanian devils and bivalves. The legitimate frame for the Hymenolepis case, and the count that disciplines it. A review in evolutionary biology.

111. 111
     Kikuchi T, Dayi M, Hunt VL, et al. Genome of the fatal tapeworm Sparganum proliferum uncovers mechanisms for cryptic life cycle and aberrant larval proliferation.Commun Biol · 2021 · 4(1):649doi:10.1038/s42003-021-02160-8 · PMID 34059788

     A tapeworm that proliferates in human tissue and is always fatal, with the proliferation attributed to extracellular matrix control and loss of sexual maturity. Genomic and transcriptomic analysis, not a study of human cases.

112. 112
     Kim TS, Pak JH, Kim JB, Bahk YY. Clonorchis sinensis, an oriental liver fluke, as a human biological agent of cholangiocarcinoma: a brief review.BMB Rep · 2016 · 49(11):590-597doi:10.5483/bmbrep.2016.49.11.109 · PMID 27418285

     The authority for three helminths in IARC Group 1, and for a national prevalence of 1.86% in Korea in 2012. A narrative review.

113. 113
     Berriman M, Haas BJ, LoVerde PT, et al. The genome of the blood fluke Schistosoma mansoni.Nature · 2009 · 460(7253):352-8doi:10.1038/nature08160 · PMID 19606141

     363 megabases and at least 11,809 genes, with more than 300 proteases: a real, gene-rich animal genome, which is why the ortholog question is answerable at all.

114. 114
     Young ND, Stroehlein AJ, Kinkar L, et al. High-quality reference genome for Clonorchis sinensis.Genomics · 2021 · 113(3):1605-1615doi:10.1016/j.ygeno.2021.03.001 · PMID 33677057

     A chromosome-level genome for a Group 1 carcinogenic fluke affecting about 35 million people. Chromosome-level contiguity is not the same as a published chromosome count, and this abstract gives no count.

115. 115
     Smout MJ, Laha T, Chaiyadet S, Brindley PJ, Loukas A. Mechanistic insights into liver-fluke-induced bile-duct cancer.Trends Parasitol · 2024 · 40(12):1183-1196doi:10.1016/j.pt.2024.10.012 · PMID 39521672

     Narrative review by the granulin discoverers: secreted growth factors, enzymes and extracellular vesicles plus chronic wounding and inflammation drive hyperplasia, fibrosis and transformation. A review, by authors with a translational stake in the molecule.

116. 116
     Dastpeyman M, Wilson DT, Loukas A, Smout MJ, Daly NL. The C-terminal region of Ov-GRN-1 granulin-1 promotes cell proliferation in the absence of regular secondary structure.Aust J Chem · 2025 · 78(11):CH25079doi:10.1071/CH25079 · PMID 42428249

     A peptide from the fluke granulin promotes cell proliferation without a folded structure, in a program aimed at topical wound-healing agents. The PubMed title has the protein name stripped; restored from the abstract's Ov-GRN-1. In vitro chemistry; cited for the group's translational interest.

117. 117
     Weglage J, Wolters F, Hehr L, et al. Schistosoma mansoni eggs induce Wnt/β-catenin signaling and activate the protooncogene c-Jun in human and hamster colon.Sci Rep · 2020 · 10(1):22373doi:10.1038/s41598-020-79450-4 · PMID 33361772

     Parasite egg antigens switch on host carcinogenic signaling, confirmed in colon biopsies from infected patients. Animal and in vitro work with human observational confirmation; the patient's own genes are what is driven.

118. 118
     Plieskatt JL, Deenonpoe R, Mulvenna JP, et al. Infection with the carcinogenic liver fluke Opisthorchis viverrini modifies intestinal and biliary microbiome.FASEB J · 2013 · 27(11):4572-84doi:10.1096/fj.13-232751 · PMID 23925654

     Fluke infection perturbs the gut and biliary microbial community in the hamster model that reliably produces cholangiocarcinoma. An animal study, and a reminder that a parasite brings its own microbiota into a sampled organ.

119. 119
     Chan-On W, Nairismägi ML, Ong CK, et al. Exome sequencing identifies distinct mutational patterns in liver fluke-related and non-infection-related bile duct cancers.Nat Genet · 2013 · 45(12):1474-8doi:10.1038/ng.2806 · PMID 24185513

     209 bile duct cancers, 108 of them fluke-related: different mutation patterns by cause, with TP53 mutations more frequent in the fluke-related group. Human tumor genes, differing by exposure.

120. 120
     Jusakul A, Cutcutache I, Yong CH, et al. Whole-Genome and Epigenomic Landscapes of Etiologically Distinct Subtypes of Cholangiocarcinoma.Cancer Discov · 2017 · 7(10):1116-1135doi:10.1158/2159-8290.CD-17-0368 · PMID 28667006

     489 cancers from 10 countries with 71 whole genomes, resolving into four clusters split by fluke exposure. The PubMed abstract has had gene symbols stripped, so none are quoted from it.

121. 121
     Hasby Saad MA, Watany MM. Schistosoma mansoni and endocarditis: from egg to free DNA detection in Egyptian patients and infected BALB/c mice.J Helminthol · 2019 · 93(2):139-148doi:10.1017/S0022149X17001183 · PMID 29352830

     Cell-free parasite DNA in the blood of 6.98% of 186 patients, and in 95% of blood and 45% of cardiac samples in infected mice. Found by a species-specific assay built to find it; the authors' line about possible genome integration is a call for research, not a finding.

122. 122
     Zhong X, Isharwal S, Naples JM, et al. Hypermethylation of genes detected in urine from Ghanaian adults with bladder pathology associated with Schistosoma haematobium infection.PLoS One · 2013 · 8(3):e59089doi:10.1371/journal.pone.0059089 · PMID 23527093

     Host gene methylation measurable in the urine of people with parasite-associated bladder damage, two markers predicting severe damage. A 57-sample pilot, host genes rather than parasite sequence.