Katherine S. Pollard

The paper introduces ChAI-Net, a deep learning model that predicts chromatin co-accessibility from DNA sequences in mouse embryonic stem cells, revealing sequence motifs that govern coordinated TF binding.

The paper uses deep learning to analyze Alu elements' impact on genome folding, identifying sequence and genomic context features that influence chromatin interactions and boundary formation.

This study shows S2F models underestimate causal variant effects, limiting their utility for fine-mapping, but top predictions are reliable.

The paper develops a computational method to predict structural variant impacts on genome folding and identifies ASD-associated variants disrupting neuronal regulatory interactions.

This paper introduces CREAM, a deep learning model for differential expression prediction using personalized genomes, showing improved performance on training genes but challenges in generalizing to unseen genes.

EukDetect2 improves eukaryote detection in metagenomes with a larger database and quantitative metrics, revealing Blastocystis and Dientamoeba as prevalent gut eukaryotes.

The paper examines genomic diversity in gut bacteria, finding that accessory genes show faster linkage decay compared to core SNPs, with mobile genetic elements showing minimal linkage to core SNPs.

Antibiotic disruption of the gut microbiome reduces heterotopic ossification in FOP mice via an IL-1 mediated pathway.

The paper uses a convolutional neural network to predict chromatin organization disruptions caused by somatic structural variants in pediatric tumors, identifying high-impact SVs and regions associated with cancer.

The paper shows human-specific enhancer HAR1984 promotes neurogenesis by altering chromatin interactions and gene regulation in cortical organoids and mice.

The paper develops a machine-learning model to predict how structural variants disrupt 3D chromatin structure in congenital heart disease and validates its predictions using induced pluripotent stem cells.

This paper develops a deep learning model called Performer to predict gene expression from personal genomes, achieving accuracy close to cis-heritability of most genes.

The paper integrates 4D Nucleome data to analyze 3D genome structure and function in human cells, revealing chromatin domains, loops, and interactions, and showing how genetic variants affect genome folding.

The paper analyzes intra-species genetic diversity of gut bacteria across individuals and links strain-level variations to host physiological traits.

The paper identifies cis-regulatory elements and transcription factors controlling FOXP3 expression in human Tregs and Tconvs, revealing mechanisms that restrict FOXP3 to Tregs in mice.

This paper develops self-supervised learning models for gut microbiome metagenomic data, showing improved prediction performance in data-limited and cross-cohort settings.

This study benchmarks genomic footprinting algorithms for predicting transcription factor binding sites in single-cell data, finding peak-level read coverage as the strongest predictor and proposing an ensemble model to improve concordance with ChIP-seq.

GAME provides a standardized framework for benchmarking genomics models across tasks, enabling consistent evaluation and community contributions.

microSLAM identifies microbiome-disease associations by analyzing gene presence/absence across strains, revealing 49 species and 83 genes linked to IBD.

The paper introduces a model that improves transcription factor binding prediction by integrating DNA sequence and structural features, enhancing interpretability and regulatory element identification.

This study introduces novel methods for comparing chromatin contact maps and evaluates their performance using Hi-C data and in silico maps, highlighting the need for biologically informed approaches.

The paper develops a method to detect invertons in gut microbiome metagenomic data and finds associations between inverton dynamics and intestinal colonization and surface adhesion.

ChromaFactor deconvolves single-molecule chromatin organization data using non-negative matrix factorization to identify key functional components and their molecular contributions.

This study shows that reduced TBX5 dosage affects 3D chromatin organization in human cardiomyocytes, with cell-to-cell variation and partial protection by CTCF co-occupancy.

The paper uses sequence-based machine learning to predict 3D genome contact maps in bonobos and chimpanzees, identifying divergent folding patterns and variants affecting genome structure.

The paper uses machine learning to infer 3D chromatin contact patterns across diverse human populations and identifies regions with significant 3D divergence not explained by sequence variation.

This study constructs pangenomes of 728 human gut prokaryotic species, revealing genetic diversity linked to stress response, antibiotic resistance, and host characteristics.

This study shows that the oxytocin receptor regulates specific aspects of pair bonding and promiscuity in prairie voles by controlling gene expression and neural mechanisms.

The paper introduces a deep learning framework, AkitaR, to study how chromatin-associated RNAs influence genome folding in HFFc6 cells, identifying known and novel RNAs involved in chromatin architecture.