Leveraging the largest harmonized epigenomic data collection for metadata prediction validated and augmented over 350,000 public epigenomic datasets
Bibliographic record
Abstract
Abstract Epigenomic data found in public databases often suffer from issues of non-standardization and incompleteness in their associated metadata. There are currently no automated approaches to validate or correct missing or inaccurate information listed in databases. To tackle this challenge, we harnessed the extensive harmonized data and metadata provided by the EpiATLAS project of the International Human Epigenome Consortium (IHEC) to train EpiClass, a suite of machine learning classifiers that can predict key metadata (∼98% accuracy), including experimental assay, donor sex, biospecimen and sample cancer status. The development of these classifiers enabled the identification of a few mislabeled and low-quality datasets in the EpiATLAS project, while also completing with high-confidence most of the missing metadata. These classifiers were also validated on ENCODE datasets absent from the initial training, then applied to assess more than 350,000 human ChIP-Seq and RNA-Seq datasets from public repositories. Overall, this effort not only validated the accuracy of the vast majority of assays reported by the original authors, but also unveiled ∼500 datasets with discrepancies, in particular through data swap within series of experiments. More importantly, EpiClass also supplied high-confidence predictions for over 320,000 metadata attributes of the biological sample such as the sex, cancer status and biomaterial type, which had been originally omitted in the majority of cases. Our work introduces the first systematic approach for metadata correction and augmentation, enhancing the quality and reliability of publicly available epigenomic data.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.012 | 0.022 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.004 | 0.004 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.002 | 0.006 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".