Abstract A006: Data Curation and Knowledge Integration Pipeline for Biomarker Discovery
Bibliographic record
Abstract
Abstract Large-scale genome sequencing data from The Cancer Genome Atlas Program (TCGA) and the International Cancer Genome Consortium (ICGC) provided rich and comprehensive catalogues of many cancer types/subtypes for a large number of donors across many different cancer types. Manually curated knowledge from databases such as Reactome (https://reactome.org), OncoKB (https://www.oncokb.org), and many others, provided complementary information on pathways, clinical relevance, and therapeutic targets that are invaluable for mining biomarkers and druggable targets. One of the challenges is to harmonize all of the data so they can be uniformly processed to identify and rank potential targets. We obtained and preprocessed data (mRNA, mutations, CNVs, protein abundance, etc.) for all 32 cancer types. We also developed an internal data portal with a web interface using the Overture portal UI (https://www.overture.bio/) that allows users to efficiently interact with our preprocessed data. We are also implementing machine learning models to classify and rank potential targets for each of the available cancer types. Our machine learning models will include predictors such as differential expression between tumor and normal, tissue specificity, co-expressions with known oncogenes/tumor suppressors, copy number variations and mutation rates. Potential predictors will be evaluated and the most relevant will be included in the final models. Using known positive and negative test cases from literature, we will evaluate, optimize, and determine the best and most relevant predictors and machine learning models to use. The selected models will then be used to prioritize and rank targets by their biomarker potential. We will manually curate our results and select candidates for validation. This project provides a harmonized data structure and ML models for searching and ranking potential biomarkers and targets in an efficient and automated way. In addition, our data portal and preprocessed data allow more efficient sharing of data and improved data accessibility and reproducibility. The use of machine learning models using the preprocessed data, combined with the data portal, additional data and literature, can result in the generation of new knowledge and advances in drug discovery. Citation Format: Samantha Majoros, Mitchell Shiell, Joe Wang, Justin Richardsson, Quang M. Trinh, Richard Marcellus, David Uehling, Rima Al-awar, Shraddha Pai, Melanie Courtot, Lincoln Stein. Data Curation and Knowledge Integration Pipeline for Biomarker Discovery [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Artificial Intelligence and Machine Learning; 2025 Jul 10-12; Montreal, QC, Canada. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(13_Suppl):Abstract nr A006.
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 | 0.005 | 0.014 |
| Meta-epidemiology (narrow) | 0.004 | 0.002 |
| Meta-epidemiology (broad) | 0.002 | 0.004 |
| Bibliometrics | 0.009 | 0.006 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.005 | 0.003 |
| Open science | 0.004 | 0.005 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.063 | 0.052 |
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".