Benchmarking Bayesian colocalization methods in validating Mendelian randomization-based target discoveries from circulating proteins for cardiometabolic diseases
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Abstract Background Mendelian randomization (MR) is an important tool for identifying potential biomarkers and drug targets. Colocalization analysis is crucial for validating MR findings and guarding against potential confounding due to linkage disequilibrium. We aim to systematically benchmark the performance of four Bayesian colocalization methods in validating MR-based target discoveries from circulating proteins for cardiometabolic diseases. Results We conducted MR analyses to assess the associations between circulating levels of 1,535 proteins and five cardiometabolic traits, followed by colocalization analyses using coloc, coloc+SuSiE, PWCoCo and SharePro. All methods demonstrated well-controlled false discoveries in the colocalization analysis of 611 pairs of circulating proteins and cardiometabolic traits with a nominal p-value > 0.9 in MR. SharePro demonstrated the highest frequency in supporting 160 (79.6%) of the 201 Bonferroni-significant protein-trait associations identified by MR, compared to coloc (supporting 40.3% of these associations), coloc+SuSiE (46.8%), and PWCoCo (45.8%), and was robust to varying prior colocalization probabilities. Moreover, protein-trait associations identified by MR and supported by SharePro were more likely to agree with significant gene-level associations based on rare variants detected in exome-wide association studies and implicate known drug targets for cardiometabolic diseases. Eight protein-trait associations were exclusively supported by SharePro and did not demonstrate a high risk of horizontal pleiotropy, suggesting potential cardiometabolic biomarkers or drug targets, such as HSF1 and HAVCR2. Conclusions SharePro most often supports high-confidence associations identified through MR for cardiometabolic diseases. Combining multiple lines of evidence using different methods may substantially increase the yield of biomarker and drug target discovery programs.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 it