Olga G. Troyanskaya

This paper introduces Seal, an AI framework that models gene regulation and variant effects in the developing human brain across multiple regions, cell types, and stages, identifying regulatory mechanisms linked to neuropsychiatric diseases.

The paper builds an InterVir Atlas to map temporal interactome changes across RNA viruses and identifies convergent host vulnerabilities for antiviral drug repurposing.

The paper generates disease-specific genetic effect maps on the blood transcriptome for five glomerular diseases using whole genome and transcriptome sequencing data from the CureGN cohort.

The paper introduces Mahi, a graph neural network framework that integrates molecular context and functional connectivity to predict gene essentiality and perturbation responses across tissues.

Volaria integrates coding and regulatory variation into gene-centered representations for disease outcome prediction using whole-genome sequencing data.

Otari predicts tissue-specific isoform abundance changes caused by genetic variants using a graph-based attention model trained on genomic and long-read transcriptome data.

Alvessa improves accuracy in functional genomics and drug target assessment by integrating evidence-based reasoning and verification.

The paper presents a multi-modal atlas of human kidney disease, identifying cellular and spatial mechanisms of injury resolution or failure and linking them to clinical outcomes.

This study uses whole-exome sequencing and single-cell DNA sequencing to analyze subclonal evolution in breast cancer during aromatase inhibitor treatment, identifying resistance mechanisms and actionable targets.

The paper analyzes the epigenomic changes in innate immune cells after influenza virus infection and finds coordinated rewiring of the epigenetic landscape.

Mixture modeling of 239 phenotype features in 5,392 SPARK autistic children defines four replicable classes carrying distinct common and rare variant signatures and developmental gene-timing patterns.

The paper uses single-nucleus multiome sequencing to analyze gene regulatory changes in human skeletal muscle after endurance exercise, identifying fiber-type specific regulatory programs and transcription factor networks.

This study identifies astrocyte-derived extracellular matrix proteins, particularly SPARC, as regulators of synapse remodeling in stress-induced depression.

SPEEDI automates single-cell multi-omics analysis by integrating data, inferring batches, and labeling cell types without user input, improving reproducibility and accessibility.

This paper introduces cGen, an unsupervised contrastive pre-training method for sequence-based genomic models, showing it improves performance in tasks like chromatin profiling and gene expression prediction.

The paper introduces CREMA, a framework for mapping cis-regulatory circuits using single-cell multiome data without peak-calling, and validates it using mouse pituitary data.

The paper identifies region-specific astrocyte genes and pathways linked to Alzheimer's disease susceptibility using a network-guided approach and mouse gene expression data.

This study shows that SARS-CoV-2 N-protein uses two distinct RNA motifs, both requiring dsRNA, to drive phase separation at specific temperatures, enabling multiple viral functions.

The paper introduces Sei, a deep learning framework that maps sequence-based regulatory activities to predict variant effects on traits and diseases using human genetics data.

The paper constructs a reference tissue atlas of the human kidney using multi-omics data to identify cell types, pathways, and genes across nephron segments and interstitium.

The paper identifies DEK as a driver of tau pathology in Alzheimer's disease vulnerable neurons using in silico modeling and experimental validation.

This paper introduces a quasilinear framework for single-cell data analysis that improves interpretability and generalizability across datasets, revealing new insights into cell lineage and neuronal populations.

This study identifies neuron-glia signaling mechanisms underlying the delayed antidepressant response to SSRIs in mice using single-cell sequencing.

DeepArk uses deep learning to model cis-regulatory codes in four model species, accurately predicting regulatory features and variant impacts.

This paper introduces AMBER, an automated framework for designing deep CNNs in genomics, showing it outperforms non-NAS and expert-designed models in accuracy.

AMBIENT is a NAS framework that creates accurate CNN architectures for genomic sequences with reduced computing costs.

The paper identifies RBP target site dysregulation as a major contributor to psychiatric disorder risk, revealing new insights into disease heritability and providing a resource for further research.

This study identifies a functional gene module associated with selective neuronal vulnerability in Alzheimer’s disease using integrated molecular and genetic data from mice and humans.

The paper uses deep learning to analyze whole-genome data from ASD families, showing noncoding mutations significantly impact transcriptional and post-transcriptional regulation and contribute to autism.

Selene is a PyTorch-based deep learning library for biological sequences, enabling model development, training, and application for various biological data.