Melissa Gymrek
Research Interests
Make your next move.
Prepare your application
Review drafts before you send them.
Explore related searches
Articles (8)
Mouse meander tail (mea) mutations produce kinked tails and selective malformation of the cerebellum anterior compartment. The anterior cerebellum defects are cell autonomous with respect to granule cell precursors, but the molecular basis has not been known. Myb-like, SWIRM, and MPN domain containing protein 1 (MYSM1) is a chromatin-associated deubiquitinase that promotes gene expression by removing monoubiquitin from histone H2A, among other targets. Loss of MYSM1 function in mice or humans results in bone marrow failure with defective maturation of B cell lineages. Here we show that extant mea alleles have mutations in Mysm1 and cause both neurological and hematological phenotypes, as do new non-complementing endonuclease-mediated mutations. Multimodal single-nucleus assays show Mysm1 effects on gene expression in several lineages and on the proportion of granule cell precursors by E14.5. Intriguingly, Mysm1 orthologs have been independently lost in several animal and fungal lineages, including yeast, flies, and nematodes. These results unite previously disconnected literature and demonstrate a requirement for MYSM1 activity in compartment-specific development of the cerebellum and suggest potential for compensatory pathways.
Year:
2026
Genetic studies have largely focused on homogeneous populations, limiting our understanding of the genetic architecture of complex traits in admixed individuals. The advent of diverse biobanks like the All of Us Research Program (AoU) and computationally efficient local ancestry inference (LAI) methods now enable admixture mapping (ADM) at biobank scale. Here, we used two orthogonal LAI methods (GNOMIX and FLARE) to characterize local ancestry in the entire AoU v7.1 cohort (n=230,019). We then used GNOMIX labels to identify associations between African (AFR) and Native American (NAT) local ancestry with 29 quantitative traits. We first analyzed All of Us v7.1 data across African ( n =49,797) and Admixed American ( n =40,327) cohorts, which identified 97 significant local ancestry associations (65 AFR, 32 NAT). These include strong known signals, such as an association between AFR ancestry at the DARC locus and white blood cell traits and between NAT ancestry at the BUD13/APOE5/ZPR1 locus and triglycerides, as well as additional signals not previously associated with local ancestry. We observed that trait associations with AFR local ancestry are largely consistent across the African and Admixed American cohorts, but that several AFR signals reach genome-wide significance exclusively in Admixed Americans. Grouping associations by trait category revealed distinct ancestral patterns: all endocrine, renal, and 75.0% of liver signals were driven by associations with NAT ancestry, whereas white blood cell (90.9%), red blood cell (65.1%), and lipid (66.7%) signals were largely associated with AFR ancestry, possibly reflecting different population-driven environmental exposures throughout history. Finally, we performed a second round of analysis, comprising the largest ADM study to date, on the entire AoU v7.1 cohort in which we pooled individuals from all ancestries. Despite evidence of confounding due to population structure, summary statistics for pooled results showed strong correlation (r>0.98) with those from single ancestry analysis and detected 2.7x fold more signals, most of which passed at least nominal significance and all of which showed consistent effect directions in the single ancestry cohorts. Overall, these results demonstrate the power of using large, admixed cohorts to gain new insights into the relationship between local ancestry and the genetic architecture of medically relevant complex traits.
Year:
2025
Adjusting for non-genetic factors can improve genetic association testing and polygenic prediction, yet most studies rely on linear adjustments for a small set of covariates. Location and time covariates are typically not included in genetic studies, but they can serve as useful proxies for environmental exposures such as climate, pollution, and behavioural differences. These features especially highlight the weakness of linear covariate adjustments, which cannot capture their cyclical or nonlinear effects nor interactions between covariates. To evaluate the benefits of modeling nonlinear covariate effects, including spatiotemporal features, we adopted a null model approach in which an auxiliary nonlinear model predicts the phenotype from non-genetic covariates alone. This prediction is then included as an additional covariate in downstream analysis, allowing association studies, polygenic scores, and fine-mapping to account for nonlinear covariate effects and interactions without modifying existing tools. Using 16 continuous phenotypes from the UK Biobank, we first evaluated gradient boosted decision tree (GBDT) and neural network null models with combinations of additional time-of-day, time-of-year, and birth or home location covariates and determined how well these models could predict phenotypes from covariates alone compared to linear models. A GBDT null improved average R 2 by 4.3% over a linear null across all phenotype-covariate set pairs, and models including spatiotemporal covariates achieved the best performance for all phenotypes. Incorporating nonlinear null predictions with additional spatiotemporal features improved polygenic scores for all phenotypes, with median improvements of 7.3% for BASIL and 15.5% for PRS-CS over the standard approach. We additionally tested three case-control phenotypes and observed improvement beyond the 95% CI of the standard approach for depression. Finally, model interpretation identified both known and novel covariate interactions and nonlinear effects. This highlighted several important cases which linear adjustments would be unable to capture, including examples where phenotype means exhibit opposite age-related trends between sexes, cyclical temporal patterns for vitamin D reflecting seasonal variation, and complex spatial dependencies between birth and home locations. Together, these results demonstrate that a small addition to existing genetic analysis workflows can improve predictive performance and provide new insights on environmental effects for complex traits.
Year:
2025
Genetic variation at thousands of short tandem repeats (STRs), which consist of consecutive repeated sequences of 1-6bp, has been statistically associated with gene expression and other molecular phenotypes in humans. However, the causality and regulatory mechanisms for most of these STRs remains unknown. Massively parallel reporter assays (MPRA) enable testing the regulatory activity of a large number of synthesized variants, but have not been applied to STRs due to experimental and computational challenges. Here, we optimized an MPRA framework based on random barcoding to study the impact of variation in repeat copy number on expression. We first performed an MPRA on sequences derived from 30,516 promoter-proximal STR loci along with up to 152bp of genomic context, testing 3-4 variants with differing repeat copy numbers for each locus in HEK293T cells. We identified 1,366 loci with significant associations between repeat copy number and expression, which were enriched for positive effect sizes (P=2.08e-110). We then designed a second MPRA in which we performed deeper perturbations, including systematic manipulation of the repeat unit sequence, orientation, and copy number, with 200-300 perturbations for each of the 300 loci with the strongest signals. Our results revealed that the repeat unit sequence is the primary driver of differences in the relationship between copy number and expression across loci, whereas orientation and flanking sequence have weaker effects, primarily for AT-rich repeat units. The high resolution of these perturbations enabled us to detect non-linear effects, most notably for AAAC/GTTT repeats, which emerge only beyond a certain copy number threshold. Finally, we observed that a subset of STRs in our library show expression levels that are tightly linked with predicted DNA secondary structure formation. We repeated our perturbation MPRA in HeLa S3 cells under wildtype and RNase H1 knockdown conditions, which, via reduction in RNase H1 activity, are expected to hinder resolution of R-loops. This demonstrated that associations between copy number and expression at G-quadruplex-forming CCCCG/CGGGG repeats are particularly sensitive to loss of RNase H1, providing support for an R-loop mediated mechanism for these repeats. Altogether, we establish STRs as a critical component of the non-coding regulatory grammar and provide a framework for understanding how this dynamic form of genetic variation shapes gene expression.
Year:
2025
Polygenic risk scores (PRSs) serve as quantitative metrics of genetic liability for various conditions. Traditionally calculated as an effect size weighted genotype summation, this formulation assumes conditional feature independence and overlooks the potential for complex interactions among genetic variants. Transformers, a class of deep learning architectures known for capturing dependencies between features, have demonstrated remarkable predictive power across domains. In this work, we introduce VADEr, a Vision Transformer (ViT)-inspired architecture that combines techniques from both natural language processing and computer vision to capture properties exhibited by genetic data and model local and global interactions for genotype-to-phenotype prediction. Evaluating VADEr's performance in predicting prostate cancer (PCa) risk, we found that across a range of metrics, including accuracy, average precision, and Matthews correlation coefficient, VADEr outperformed all benchmark methods, demonstrating its effectiveness in the context of complex disease risk prediction. To illuminate identified drivers of disease risk by VADEr, we formulated DARTH scores, an attention-based attribution metric, to capture the personalized contribution of each genomic region. These scores revealed distinct genetic heterogeneity captured by VADEr, with drivers of predicted risk identified in key PCa risk regions including the HOXB13, TMPRSS2, and MSMB loci. DARTH scores also revealed germline predispositions for particular PCa molecular subtypes, including an association between the LMTK2 locus and the SPOP subtype, both implicated in the regulation of androgen receptor activity. Overall, by effectively capturing dependencies among genetic variants and providing interpretable insights, VADEr and DARTH scores offer a promising direction for advancing genotype-to-phenotype prediction, particularly in complex disease.
Year:
2025
Tandem repeats (TRs) - highly polymorphic, repetitive sequences across the human genome - are important regulators of gene expression but remain underexplored due to challenges in accurate genotyping and analysis. Here, we generate new whole genome and single-cell RNA sequencing from >5.4 million blood-derived cells across 1,925 individuals in two cohorts, and perform meta-analysis to characterize the impact of variation in >1.7 million TR loci on immune cell type-specific gene expression. We identify >69,000 single-cell expression TR loci (sc-eTRs), 30.7% of which are specific to one of 28 immune cell types, and reveal dynamic regulatory effects using cell-state inference. Matched single-cell ATAC-seq profiles from >3.4 million nuclei in 922 individuals uncover chromatin accessibility QTLs for nearly one-third of expression-associated TRs, supporting coordinated effects on cis-regulatory architecture. Fine-mapping implicates 1,490 TRs as candidate causal drivers of gene expression in 6.1% of tested genes, and colocalization analyses highlight >200 genes in which TRs likely mediate genetic associations with immune and hematological traits. Together, these results provide a genome-wide, multiomic view of TR-mediated regulation in the human immune system, establishing TRs as key contributors to cell type-specific regulatory variation and complex trait architecture.
Year:
2024
Motivation Somatic mosaicism has been implicated in several developmental disorders, cancers, and other diseases. Short tandem repeats (STRs) consist of repeated sequences of 1–6 bp and comprise >1 million loci in the human genome. Somatic mosaicism at STRs is known to play a key role in the pathogenicity of loci implicated in repeat expansion disorders and is highly prevalent in cancers exhibiting microsatellite instability. While a variety of tools have been developed to genotype germline variation at STRs, a method for systematically identifying mosaic STRs is lacking. Results We introduce prancSTR, a novel method for detecting mosaic STRs from individual high-throughput sequencing datasets. prancSTR is designed to detect loci characterized by a single high-frequency mosaic allele, but can also detect loci with multiple mosaic alleles. Unlike many existing mosaicism detection methods for other variant types, prancSTR does not require a matched control sample as input. We show that prancSTR accurately identifies mosaic STRs in simulated data, demonstrate its feasibility by identifying candidate mosaic STRs in Illumina whole genome sequencing data derived from lymphoblastoid cell lines for individuals sequenced by the 1000 Genomes Project, and evaluate the use of prancSTR on Element and PacBio data. In addition to prancSTR, we present simTR, a novel simulation framework which simulates raw sequencing reads with realistic error profiles at STRs. Availability and implementation prancSTR and simTR are freely available at https://github.com/gymrek-lab/trtools. Detailed documentation is available at https://trtools.readthedocs.io/.
Year:
2024
Summary Leveraging local ancestry and haplotype information in genome-wide association studies and downstream analyses can improve the utility of genomics for individuals from diverse and recently admixed ancestries. However, most existing simulation, visualization and variant analysis frameworks are based on variant-level analysis and do not automatically handle these features. We present haptools, an open-source toolkit for performing local ancestry aware and haplotype-based analysis of complex traits. Haptools supports fast simulation of admixed genomes, visualization of admixture tracks, simulation of haplotype- and local ancestry-specific phenotype effects and a variety of file operations and statistics computed in a haplotype-aware manner. Availability and implementation Haptools is freely available at https://github.com/cast-genomics/haptools. Documentation Detailed documentation is available at https://haptools.readthedocs.io. Supplementary information Supplementary data are available at Bioinformatics online.
Year:
2023
Collaborators (16)
Nathan Zemke
Assistant Professor
University of California, San Diego
Daniel Goldowitz
Professor
The University of British Columbia
Alex Hewitt
University of Tasmania
Christopher Benner
Associate Professor
University of California, San Diego
Joseph Powell
University of New South Wales
Owen Siggs
Garvan Institute of Medical Research
Gemma A Figtree
University of Sydney
Tyler Seibert
Assistant Professor
University of California, San Diego
Angli Xue
University of New South Wales
Chris Wallace
Director of Research
University of Cambridge
Bruce Hamilton
Professor
University of California, San Diego
Michael Talkowski
Professor
Massachusetts General Hospital
Hannah Carter
Professor
University of California, San Diego
Brent Rose
Associate Professor
UC San Diego Health System
Harriet Dashnow
Assistant Professor
University of Colorado Anschutz Medical Campus
Max Chang
University of California, San Diego
Make your next move.
Prepare your application
Review drafts before you send them.