#38 · Genomics
Causal and non-additive models resolve plasma protein genetic effects
The dominant thread is a move from marginal additive association toward models that partition genetic architecture into non-additive and causal components. EIR-auto-GP detects dominance and epistasis in UK Biobank plasma pQTLs, with replication in Olink and mass-spec cohorts. A causal network framework using cis-pQTL instruments on 7523 AGES serum proteins identifies 185 protein subnetworks linked to future myocardial infarction. The aortic stenosis study is a more heterogeneous strand: it applies multi-ancestry GWAS of 86,864 cases, TWAS, a polygenic risk score and CMKLR1/LTBP4 silencing validation to therapeutic target discovery, without modelling non-additive effects. What is new is explicit causal and non-additive modelling of protein and disease traits at biobank scale.
Published papers
Circulating causal protein networks linked to future risk of myocardial infarction.
Causal network inference framework using cis-pQTL instruments on 7523 serum proteins in the AGES cohort, identifying 185 protein subnetworks linked to future…
Complex genetic effects linked to plasma protein abundance in the UK Biobank.
Deep learning method EIR-auto-GP detects dominance and epistasis in plasma pQTLs in UK Biobank, replicated in Olink and mass-spec cohorts.
Genomic and transcriptomic analyses of aortic stenosis enhance therapeutic target discovery and disease prediction.
Multi-ancestry GWAS of 86,864 aortic stenosis cases yields 241 autosomal loci, 54 TWAS genes, a PRS, and CMKLR1/LTBP4 silencing validation.
Topic history
| Date | Rank |
|---|---|
| #38 | |
| #36 | |
| #30 | |
| #11 | |
| #24 | |
| #8 | |
| #36 | |
| #38 | |
| #19 | |
| #2 | |
| #17 | |
| #15 | |
| #6 | |
| #27 | |
| #13 | |
| #16 | |
| #25 | |
| #5 | |
| #13 | |
| #9 | |
| #3 | |
| #11 | |
| #15 | |
| #16 | |
| #9 | |
| #34 | |
| #29 | |
| #27 | |
| #8 | |
| #33 | |
| #5 | |
| #1 |