Zachary Ryan McCaw

Large Impact of Genetic Data Processing Steps on Stability and Reproducibility of Set-Based Analyses in Genome-Wide Association Studies

This paper shows how data processing steps in GWAS affect reproducibility and stability of set-based analyses, and provides tools to address these issues.

Functional dissection of complex and molecular trait variants at single nucleotide resolution

This study functionally characterizes 221,412 trait-associated variants using MPRA in 5 cell types, identifying 12,025 regulatory variants and revealing widespread epistasis and non-TF mechanisms.

MultiSuSiE improves multi-ancestry fine-mapping in All of Us whole-genome sequencing data

MultiSuSiE improves multi-ancestry fine-mapping in All of Us whole-genome sequencing data by allowing causal effect sizes to vary across ancestries, achieving higher power and lower computational costs.

A Scalable Framework for Identifying Allelic Series from Summary Statistics

The paper introduces COAST-SS, a method to identify allelic series using summary statistics, showing it is effective and robust to LD matrix misspecification.

A statistical framework for powerful multi-trait rare variant analysis in large-scale whole-genome sequencing studies

The paper introduces MultiSTAAR, a statistical framework for multi-trait rare variant analysis in WGS studies, identifying new lipid trait associations in diverse populations.

Pitfalls in performing genome-wide association studies on ratio traits

This paper shows that genome-wide association studies on ratio traits can be misleading, as associations may be driven by the denominator rather than the numerator, and proposes a method to correct for collider bias.

Causal considerations can determine the utility of machine learning assisted GWAS

This study shows that machine learning-derived phenotypes for GWAS can lead to false discoveries depending on their causal relationship with the target phenotype and environment.

Unsupervised representation learning improves genomic discovery and risk prediction for respiratory and circulatory functions and diseases

This paper introduces a deep learning framework, REGLE, that improves genomic discovery and risk prediction for respiratory and circulatory diseases by learning low-dimensional embeddings from high-dimensional clinical data.

Leveraging a machine learning derived surrogate phenotype to improve power for genome-wide association studies of partially missing phenotypes in population biobanks

This paper introduces SynSurr, a method that improves GWAS power for partially missing phenotypes by jointly analyzing the target phenotype and a synthetic surrogate predicted by machine learning, without requiring a correctly specified imputation model.

Utilizing multimodal AI to improve genetic analyses of cardiovascular traits

This paper introduces M-REGLE, a multimodal deep learning method that improves genetic analysis of cardiovascular traits by integrating multiple clinical data modalities and outperforms unimodal approaches in locus discovery and risk prediction.

EmbedGEM: A framework to evaluate the utility of embeddings for genetic discovery

EmbedGEM evaluates embeddings for genetic discovery by assessing heritability and disease relevance using polygenic risk scores and genome-wide association statistics.

High-throughput ML-guided design of diverse single-domain antibodies against SARS-CoV-2

This study uses ML to design diverse VHH antibodies that neutralize SARS-CoV-1 and SARS-CoV-2 variants, including Delta and Omicron BA.1.

An allelic series rare variant association test for candidate gene discovery

The paper introduces COAST, a rare variant association test for allelic series, which improves power over SKAT-O in detecting gene-trait associations in lipid and cell count traits using UK Biobank data.

ChoruMM: a versatile multi-components mixed model for bacterial-GWAS

The paper develops ChoruMM, a multi-components mixed model for bacterial GWAS, showing that standard human genetics models are inadequate for highly structured bacterial genomes.

Leveraging deep-learning on raw spirograms to improve genetic understanding and risk scoring of COPD despite noisy labels

This paper trains a deep neural network on noisy COPD labels to predict COPD status from spirograms, improving genetic discovery by identifying new loci and replicating existing ones.

MGMM: An R Package for fitting Gaussian Mixture Models on Incomplete Data

MGMM fits Gaussian mixture models on incomplete data, outperforming imputation methods in recovering cluster assignments and assessing uncertainty.

Leveraging a surrogate outcome to improve inference on a partially missing target outcome

The paper introduces a method to improve eQTL inference in undersampled tissues by jointly modeling a target outcome and a correlated surrogate outcome.

DeepNull: Modeling non-linear covariate effects improves phenotype prediction and association power

Multitrait genetic-phenotype associations to connect disease variants and biological mechanisms

The paper decomposes multitrait genetic associations to identify pleiotropy profiles linking SNPs to biological pathways and drug targets.

Operating Characteristics of the Rank-Based Inverse Normal Transformation for Quantitative Trait Analysis in Genome-Wide Association Studies

This paper defines and compares two INT-based association tests for non-normally distributed traits in GWAS, introduces an adaptive omnibus test, and shows it outperforms standard tests in power and type I error control.