Suyash Shringarpure

The paper introduces distributed AI agents to analyze large-scale biological datasets, demonstrating their ability to rapidly extract insights and predict outcomes across a Perturb-seq atlas.

LifeSciBench evaluates language models on 750 expert-authored tasks across seven life science domains and workflows, revealing significant gaps in model performance.

This paper explores the use of LLM-based coding agents in scientific computing, presenting case studies in the life sciences and discussing their potential and challenges.

The paper analyzes the impact of CYP2C19 genotype on SSRI side effects and discontinuation rates in a large sample of individuals.

GeneBench-Pro evaluates AI agents' ability to perform multi-stage scientific analyses in genomics and biomedicine, showing improved benchmarking over previous versions with mixed model performance.

This paper develops a polygenic risk score for glaucoma with improved performance across major ancestries, enabling earlier detection and targeted treatment of high-risk individuals.

This study finds that higher polygenic risk scores for breast cancer are associated with longer survival in patients, suggesting a link between genetic predisposition and less aggressive tumors.

This study uses a multivariate GWAS approach to identify genetic loci associated with reading ability and dyslexia, finding 35 new genome-wide significant loci and estimating SNP heritability at 24%.

This study compares polygenic risk scores for Parkinson's disease across seven ancestries, finding varying predictive accuracy and highlighting the need for ancestry-specific models.

PRSformer predicts disease risk from large-scale genotypes using deep learning, outperforming linear models and summary-statistic methods in multi-ancestry cohorts.

This study compares eQTL discovery in African and European ancestry cohorts, finding more eQTLs in African samples and improved GWAS variant-to-gene mapping using African eQTLs.

This study identifies genetic correlations between hyposmia and Parkinson's disease, finding a strong causal link from PD to hyposmia but weaker evidence for the reverse.

This study investigates the shared genetic architecture between rhythm and dyslexia using large-scale genome-wide association studies.

This paper evaluates how the scale and quality of genetic data from the 23andMe cohort improve predictions of clinical success for drug targets.

This study shows large language models can accurately prioritize causal genes at GWAS loci, outperforming existing methods and offering a scalable solution.

MUSSEL improves polygenic risk prediction across populations by integrating GWAS summary statistics and using Bayesian hierarchical modeling with ensemble learning.

The paper uses the 23andMe database to study MAP3K15's metabolic role, finding LoF carriers have delayed T2D diagnosis and reduced cholesterol.

This paper conducts a multi-ancestry genome-wide meta-analysis of Parkinson's disease, identifying 78 independent loci and 23 genes associated with PD risk.

This paper uses a polygenic risk score to identify undiagnosed diabetes cases, finding that combining BMI and PRS captures most undiagnosed cases, including those with normal weight.

The paper estimates polygenic risk scores (PRS) for 12 diseases in a diverse cohort and finds higher disease incidence and earlier clinical manifestation in top PRS individuals.

The paper introduces CT-SLEB, a method for multi-ancestry polygenic prediction that improves performance in non-European populations using GWAS summary statistics and simulation studies.

This study identifies 49 genetic loci associated with sleep apnoea risk using multi-trait GWAS, with 29 replicated in a large cohort, and finds genetic correlations with several complex traits.

This study conducts a genome-wide association study to identify genetic variants linked to musical beat synchronization, finding high polygenicity with 69 significant loci and heritability estimates of 13%-16%.

The paper constructs polygenic indexes for 47 phenotypes across 11 datasets and provides a tool to correct errors-in-variables bias when using PGIs as proxies for the additive SNP factor.

This paper uses self-reported medical data to perform GWAS on 33 rare diseases, rediscovering known associations and finding novel ones, validated across multiple populations.

This study identifies 178 independent risk loci for depression through a large meta-analysis across multiple cohorts, including fine-mapping and transcriptome-wide association findings.

The paper identifies sex-differential participation bias in genetic studies by performing GWAS on sex in 3.3 million individuals, finding 158 loci associated with sex and highlighting traits like BMI that may influence participation.

The paper identifies novel genetic risk loci for depression and anxiety using factor analysis and genomic SEM, revealing shared and trait-specific genetic influences.

This study identifies genetic loci associated with daytime napping and finds potential causal links between napping and cardiometabolic traits.

This paper introduces TPBWT for fast and robust IBD inference, showing improved speed and accuracy over existing methods, especially in handling phasing errors.