Toward generalizable prediction of cancer signal using a cell-free DNA language model
Cell Reports Medicine, published 2026-06-12; Europe PMC full-text JATS XML of PMC13400139 · Peer-reviewed
DOI: 10.1016/j.xcrm.2026.102866
Developer study: ten authors are employees of Nanjing Geneseeq Technology Inc., which sequenced the samples and developed Fragmentia-AI. The baseline feature pipelines were re-implemented by the authors.
What this source contributed
Records in this release that cite this source. A record can cite several sources.
Results
Values as the source printed them. Compare results only under compatible protocols, datasets and metrics.
Claims
Statements this source supports, each with where in the source it was checked.
- Evidence summary: plasma ctDNA fragmentomics workflows: Summary Eight sources compare plasma cfDNA fragmentomic workflows on retrospective case-control cohorts. None is a prospective screening study and none measures clinical benefit. Cristiano et al. 2019, the DELFI developers, report 73% sensitivity (152 of 208 cancers) at a reported 98% specificity for DELFI in repeated cross-validation against 215 healthy individuals. Hou et al. 2024 re-implemented ten published fragmentation patterns and DELFI with one support vector machine each on the same cohort. They are independent of every pattern except IFS, which their corresponding author developed. In 10 x 10-fold cross-validation, pan-cancer AUC within open chromatin regions ranged from 0.8741 (fragment length) to 0.9736 (5' end motif), and their DELFI re-implementation reached 0.9575. Rankings changed on independent cohorts. On 46 lung cancers and 385 healthy individuals, with a model trained on 12 lung cancers, the end-motif model fell to AUC 0.5355 and sensitivity 0.0652 at 95% specificity, while fragment size distribution reached 0.9386 and 0.7174. Hou et al.'s supplementary pan-cancer and liver cancer sensitivities conflict with other tables of the same file and are not shown. Wang et al. 2026, the UNITE developers, stratified cancers by ichorCNA tumour fraction in pooled public and new 0.1x data that include samples of the DELFI and Jiang et al. cohorts used above. For cancers with an estimated tumour fraction of 3% or less, cross-validated AUC was 0.878 for all five features combined, 0.873 for fragment length, 0.857 for the SD of length counts, 0.736 for copy number, 0.715 for the C/T end-motif ratio and 0.651 for the short/long ratio, against 0.603 for a classifier using the ichorCNA tumour fraction alone, which detected 0.058 of these cancers at 95% specificity. For cancers with an estimated fraction above 3% and up to 10%, the ichorCNA classifier reached 0.983, above the combined model (0.962) and copy number alone (0.926). The extracted UNITE tables give sensitivity at a fixed specificity only for the ichorCNA comparator; the fragmentomic models' values are in a sheet not yet extracted. Jamshidi et al. 2022 (Cancer Cell), by the GRAIL developers, scored fragment length and fragment endpoint classifiers from 30x cfDNA whole-genome sequencing on the same CCGA substudy 1 participants as their methylation, copy-number and variant classifiers, at a 98% specificity threshold set post hoc within each set. On the validation set of 464 cancers and 362 non-cancers the fragment length classifier detected 29% (95% CI 25% to 34%) and the fragment endpoint classifier 18% (15% to 22%), against 34% for whole-genome methylation analysed double-blinded; both fragment classifiers were developed after the validation set had been unblinded, so these are not blinded validation results, and no tumour-fraction-stratified sensitivity is printed. Zeng et al. 2026 (Nature Cancer), by the cfMeDIP-seq developers, computed fragmentomic features from cfMeDIP-seq reads, which are enriched for methylated fragments rather than drawn from whole-genome sequencing, and report means over 10 fold models. On an independent paired-end validation set, a fragment-ratio classifier reached AUC 0.848 and detected 0.533 of cancers at 99% specificity, and fragment ratios combined with methylation reached 0.954 and 0.766. The validation set has 17 healthy controls, so its 99% specificity allows no false positive among them, and Li-Fraumeni syndrome carriers without current cancer count as cancers in its sensitivities. Three of the five classifiers that include end motifs scored below AUC 0.5 there, which the authors attribute to healthy controls processed with a different laboratory protocol, so validation results for all five are not shown. Helzer et al. 2025 (Nature Communications) compared twelve fragmentomic metrics adapted to targeted exon panels, including the authors' own fragment-size entropy metrics, their combination and Griffin, with one classifier and shared folds, on plasma from patients with metastatic cancer. The healthy-versus-cancer one-vs-rest AUROC ranged from 0.835 (open-chromatin entropy) to 0.989 (first-exon depth) in a UW cohort of 404 cancers and 27 healthy donors, with Griffin at 0.944, and from 0.951 to 1.000 in public GRAIL panel data from 151 cancers and 47 healthy donors. In a developer comparison on 0.1x whole-genome sequencing (Cell Reports Medicine 2026), a fragment-level language model (Fragmentia-AI) reached AUC 0.9561 on a held-out set of unstated size, against 0.9209 to 0.935 for the developers' re-implementation of DELFI features and 0.881 to 0.8863 for Griffin features under three classifiers. The same company's whole-genome model (Molecular Biomedicine 2026), on an independent test cohort of 2,952 cancers and 3,044 non-cancer controls, reached AUC 0.910 for cancers with an ichorCNA tumour fraction below 3% and detected 0.46 of them at a cutoff set to 98% specificity in training (observed test specificity 0.97); among those low-fraction cancers it detected 0.391 at stage I/II (637 of 1,628) and 0.708 at stage III/IV (345 of 487). Its tumour-fraction strata are ichorCNA estimates, no comparator was run on its cohorts, and its training cohort has the same number of non-cancer controls (1,245) as the Cell Reports Medicine cohort, so the two studies may share samples. Three of these four studies report on their authors' own method, and in the fourth the authors' own entropy metrics are among those compared. Cohorts, feature implementations, classifiers and validation designs differ, so no value is comparable across sources. (amp-20261007-ctdna-fragmentomics-source: result amp-20261007-ctdna-fragmentomics-result; ctdnafrag-20261009-source-hou2024: Table 1, P24 and P44; ctdnafrag-20261009-source-hou2024-supporting-information: Table S3; ctdnafrag-20261009-source-wang2026-data-file-s2: STATS_xgb_x1-x6 and STATS_lr; ctdnafrag-20261009-source-wang2026: Results and Methods; ctdnajam-20261010-source-jamshidi2022: Table 3 validation column and Results, cancer signal origin prediction; ctdnalit-20261011-zeng2026-source-article: Results P28-P33; ctdnalit-20261011-zeng2026-source-source-data-fig4: sheets Fig.4c and Fig.4d; ctdnalit-20261011-helzer2025-source-supplementary-information: Supplementary Tables S1 and S2, 'Normal' columns; ctdnalit-20261011-helzer2025-source-article: Results P5-P8; ctdnalit-20261011-fragmentia2026-source-document-s1: Table S5; ctdnalit-20261011-fragmentia2026-source-article: Results P39-P44; ctdnalit-20261011-fragmentiawgs2026-source-article: Results P8-P21; ctdnalit-20261011-fragmentiawgs2026-source-supplementary-material-1: Tables S1 and S4)
- Relevance of ctdnalit-20261011-fragmentia2026-protocol-ulp-wgs-heldout-auc to "Which fragmentomics workflow detects tumour-derived plasma DNA under realistic low tumour fractions and fixed false-positive constraints?": Assessed by Fragmentia-AI paper: ULP-WGS cancer vs non-cancer AUC on a held-out validation set, six feature sets x three classifiers (ctdnalit-20261011-fragmentia2026-source-article: Results P39-P44; STAR Methods P80-P94; ctdnalit-20261011-fragmentia2026-source-document-s1: Table S5; Table S4)
Models, benchmarks and other records
Methods (5)
Configurations (19)
- DELFI features with Deep Learning (Fragmentia-AI paper re-implementation)
- DELFI features with Random Forest (Fragmentia-AI paper re-implementation)
- DELFI features with XGBoost (Fragmentia-AI paper re-implementation)
- FDI features with Deep Learning (Fragmentia-AI paper re-implementation)
- FDI features with Random Forest (Fragmentia-AI paper re-implementation)
- FDI features with XGBoost (Fragmentia-AI paper re-implementation)
- FRAGMA features with Deep Learning (Fragmentia-AI paper re-implementation)
- FRAGMA features with Random Forest (Fragmentia-AI paper re-implementation)
- FRAGMA features with XGBoost (Fragmentia-AI paper re-implementation)
- Fragmentia-AI trained on ULP-WGS (separate instance)
- Griffin features with Deep Learning (Fragmentia-AI paper re-implementation)
- Griffin features with Random Forest (Fragmentia-AI paper re-implementation)
- Griffin features with XGBoost (Fragmentia-AI paper re-implementation)
- Nucleosome features with Deep Learning (Fragmentia-AI paper re-implementation)
- Nucleosome features with Random Forest (Fragmentia-AI paper re-implementation)
- Nucleosome features with XGBoost (Fragmentia-AI paper re-implementation)
- Subnucleosome features with Deep Learning (Fragmentia-AI paper re-implementation)
- Subnucleosome features with Random Forest (Fragmentia-AI paper re-implementation)
- Subnucleosome features with XGBoost (Fragmentia-AI paper re-implementation)
Protocols (1)
Datasets (1)
Evaluations (19)
- cfDNA sequence / Fragmentia-AI on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- DELFI / Deep Learning on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- DELFI / Random Forest on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- DELFI / XGBoost on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- FDI / Deep Learning on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- FDI / Random Forest on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- FDI / XGBoost on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- FRAGMA / Deep Learning on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- FRAGMA / Random Forest on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- FRAGMA / XGBoost on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- Griffin / Deep Learning on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- Griffin / Random Forest on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- Griffin / XGBoost on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- Nucleosome / Deep Learning on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- Nucleosome / Random Forest on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- Nucleosome / XGBoost on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- Subnucleosome / Deep Learning on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- Subnucleosome / Random Forest on the ULP-WGS held-out validation set (Fragmentia-AI paper)
- Subnucleosome / XGBoost on the ULP-WGS held-out validation set (Fragmentia-AI paper)
Evidence
Source checking verifies the cited claim or transcription. It does not establish independent reproduction.
Evidence table
Inspect claims, sources and review details
Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.
One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.
20 evidence rows matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
| Archive note Not archived in the batch: CC BY-NC-ND 4.0. The artifact_sha256 pins the bytes that were read. Source metadata | Toward generalizable prediction of cancer signal using a cell-free DNA language model No field-specific location recorded Version: Cell Reports Medicine, published 2026-06-12; Europe PMC full-text JATS XML of PMC13400139 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Artifact SHA-256 42a96f1b59e000d55046eeaa86035b19502b803a3d826fb2f791ee597df5fdbd Source metadata | Toward generalizable prediction of cancer signal using a cell-free DNA language model No field-specific location recorded Version: Cell Reports Medicine, published 2026-06-12; Europe PMC full-text JATS XML of PMC13400139 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Artifact url https://www.ebi.ac.uk/europepmc/webservices/rest/PMC13400139/fullTextXML Source metadata | Toward generalizable prediction of cancer signal using a cell-free DNA language model No field-specific location recorded Version: Cell Reports Medicine, published 2026-06-12; Europe PMC full-text JATS XML of PMC13400139 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Doi 10.1016/j.xcrm.2026.102866 Source metadata | Toward generalizable prediction of cancer signal using a cell-free DNA language model No field-specific location recorded Version: Cell Reports Medicine, published 2026-06-12; Europe PMC full-text JATS XML of PMC13400139 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Licence CC-BY-NC-ND-4.0 Source metadata | Toward generalizable prediction of cancer signal using a cell-free DNA language model No field-specific location recorded Version: Cell Reports Medicine, published 2026-06-12; Europe PMC full-text JATS XML of PMC13400139 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Media type application/xml Source metadata | Toward generalizable prediction of cancer signal using a cell-free DNA language model No field-specific location recorded Version: Cell Reports Medicine, published 2026-06-12; Europe PMC full-text JATS XML of PMC13400139 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Publication status peer_reviewed Source metadata | Toward generalizable prediction of cancer signal using a cell-free DNA language model No field-specific location recorded Version: Cell Reports Medicine, published 2026-06-12; Europe PMC full-text JATS XML of PMC13400139 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Retrieved at 2026-10-11T21:41:11Z Source metadata | Toward generalizable prediction of cancer signal using a cell-free DNA language model No field-specific location recorded Version: Cell Reports Medicine, published 2026-06-12; Europe PMC full-text JATS XML of PMC13400139 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Scope note Developer study: ten authors are employees of Nanjing Geneseeq Technology Inc., which sequenced the samples and developed Fragmentia-AI. The baseline feature pipelines were re-implemented by the authors. Source metadata | Toward generalizable prediction of cancer signal using a cell-free DNA language model No field-specific location recorded Version: Cell Reports Medicine, published 2026-06-12; Europe PMC full-text JATS XML of PMC13400139 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Source location Full artifact bytes; per-record locators on each record. 'Pn' is the zero-based index of the <p> element in document order in this JATS XML, front matter included. Source metadata | Toward generalizable prediction of cancer signal using a cell-free DNA language model No field-specific location recorded Version: Cell Reports Medicine, published 2026-06-12; Europe PMC full-text JATS XML of PMC13400139 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
Sources and history
Release 2026-11-21-c8bea25e5abb · Record review: source checked
Primary source retrieved and hashed for the 2026-10-11 plasma ctDNA literature pass.
Technical metadata and extraction receipts
Stable ID: ctdnalit-20261011-fragmentia2026-source-article
- areas
- dna-genomes
- contexts
- clinical_research
- url
- https://doi.org/10.1016/j.xcrm.2026.102866
- artifact url
- https://www.ebi.ac.uk/europepmc/webservices/rest/PMC13400139/fullTextXML
- version
- Cell Reports Medicine, published 2026-06-12; Europe PMC full-text JATS XML of PMC13400139
- retrieved at
- 2026-10-11T21:41:11Z
- artifact sha256
- 42a96f1b59e000d55046eeaa86035b19502b803a3d826fb2f791ee597df5fdbd
- doi
- 10.1016/j.xcrm.2026.102866
- publication status
- peer_reviewed
- licence
- CC-BY-NC-ND-4.0
- media type
- application/xml
- source locator
- Full artifact bytes; per-record locators on each record. 'Pn' is the zero-based index of the <p> element in document order in this JATS XML, front matter included.
- archive note
- Not archived in the batch: CC BY-NC-ND 4.0. The artifact_sha256 pins the bytes that were read.
- scope note
- Developer study: ten authors are employees of Nanjing Geneseeq Technology Inc., which sequenced the samples and developed Fragmentia-AI. The baseline feature pipelines were re-implemented by the authors.