Research

Computation-driven discovery for biomedical science

The Brylinski Lab develops computational methods that connect molecular structure, biological networks, and biomedical data to understand disease mechanisms and guide therapeutic discovery. Our work combines domain science in biochemistry, cancer biology, microbiome research, infectious disease, and drug discovery with machine learning, artificial intelligence, data science, and high-performance computing.

Foundations Research programs How we work Computing

Research Foundations

Biomedical questions powered by computation

Our research is organized around two tightly connected foundations: the biological problems we study and the computational methods we build to study them.

Research Programs

Integrated research directions

Each program combines a biomedical challenge with a computational strategy, producing models, datasets, predictions, and mechanistic hypotheses that can guide future experimental or translational work.

01 Cancer AI

AI-guided precision oncology and drug response prediction

We develop models to predict how cancer cells respond to therapy, prioritize kinase targets, and identify promising anticancer drug combinations.

Biomedical focus Cancer biology, kinase inhibitors, drug synergy, precision therapy.
Computational strategy Graph neural networks, multi-omics integration, explainable AI, data augmentation.
Precision oncology Drug synergy Drug response
02 Networks

Graph learning for molecular and disease networks

We represent biological systems as networks to study relationships among proteins, drugs, genes, pathways, diseases, and cellular phenotypes.

Biomedical focus Protein interaction networks, disease mechanisms, cancer-specific graphs.
Computational strategy GCNs, graph embeddings, biological feature integration, graph reduction.
Graph neural networks Network medicine Multi-omics networks
03 Drug Discovery

Structure-based modeling and computer-aided drug discovery

We use molecular structure to understand ligand binding, compare binding pockets, model drug-target interactions, and support drug discovery and repurposing.

Biomedical focus Therapeutic targets, ligand binding, allosteric sites, drug repurposing.
Computational strategy Docking, pocket comparison, virtual screening, structural bioinformatics.
Drug discovery Ligand binding Structure-based modeling
04 Proteins

Representation learning for proteins, enzymes, and metabolism

We develop machine-learning representations for biochemical entities such as enzymes, metabolites, proteins, and reactions to improve prediction of biological function.

Biomedical focus Enzyme function, metabolism, microbiome, biochemical annotation.
Computational strategy Embeddings, autoencoders, CNNs, feature engineering, predictive modeling.
Protein function Enzyme function Microbiome metabolism Representation learning
05 Infectious Disease

Computational discovery of antiviral and antimicrobial agents

We apply computational modeling and AI to identify therapeutic opportunities against viral and bacterial pathogens, including drug repurposing and engineered antibacterial proteins.

Biomedical focus SARS-CoV-2, antimicrobial resistance, phage endolysins, emerging pathogens.
Computational strategy Network-based prediction, molecular modeling, docking, machine-learning scoring.
Antiviral discovery Antimicrobial discovery Herpesvirus research
06 Data Science

Large-scale biomedical and institutional data science

We build predictive models and data-driven workflows for complex datasets, emphasizing robust preprocessing, dimensionality reduction, validation, and interpretable outcomes.

Domain focus Biomedical datasets, institutional data, student success, repository-scale analysis.
Computational strategy Autoencoders, dimensionality reduction, predictive analytics, data integration.
Data science Autoencoders Educational analytics Data infrastructure

How We Work

From biomedical question to computational insight

Our projects typically move from a biological question to data integration, model development, mechanistic interpretation, and validation through literature, benchmark datasets, collaborators, or experimental follow-up.

01

Biomedical question

Define a disease, molecular mechanism, therapeutic target, or biochemical problem.

→
02

Data and structures

Integrate omics data, molecular structures, networks, annotations, and experimental evidence.

→
03

AI/ML model

Build predictive models using graph learning, embeddings, autoencoders, docking, or statistical learning.

→
04

Interpretation

Analyze predictions to identify mechanisms, important features, pathways, or molecular interactions.

→
05

Validation

Benchmark against independent data and connect predictions to literature or experimental testing.

Computational Infrastructure

Scalable computing for biomedical discovery

Many projects require large-scale computation, including network construction, docking, model training, inference across large biomedical datasets, and reproducible pipelines.

High-performance computing

Clusters, parallel workflows, and large-scale computation help us handle biomedical datasets, molecular libraries, and graph-based models that are too large for manual analysis.

Accelerators and GPUs

GPU-accelerated workflows support deep learning, graph neural networks, molecular modeling, and high-throughput calculations.

Reproducible pipelines

Structured datasets, documented workflows, reusable code, and robust evaluation help make computational results inspectable, extensible, and validatable.