
We use tools from AI and computational and mathematical modeling to solve a variety of problems in fundamental and disease biology.
A major focus of our work is the application of computational methods to study the signaling and transcriptional regulatory networks that underlie the determination of cell fate, and the perturbation of these networks under disease conditions or chemical perturbations. We are integrating diverse genomic data sets, focusing on single cell and spatial transcriptomics, to predict perturbations in cell fate and gene regulatory networks. We rely primarily on mathematical and computational modeling as research tools, and work in close collaboration with experimental scientists.
Modeling perturbations in single-cell gene expression
The space of possible human cell types and chemical treatments / doses is exponentially large. We are developing deep learning-based models to extrapolate chemical dose-response across cell types, doses and chemicals. We are also interested in the broader theoretical questions underlying developmental and disease biology, and the potential for dynamical systems theory to explain celll-level biological data.
Recent Publications:
- Dynamical systems theory as an organizing principle for single-cell biology, Islam and Bhattacharya, 2025.
- Protocol for predicting single- and multiple-dose-dependent gene expression using deep generative learning, Bowman et al, 2025.
- Unique challenges and best practices for single cell transcriptomic analysis in toxicology, Filipovic et al, 2024.
- Generative Modeling of Single Cell Gene Expression for Dose-Dependent Chemical Perturbations, Kana et al, 2023.
Predictive modeling of cellular transitions in health and disease
We are integrating and analyzing bulk and single-cell gene expression data vestigating the role of perivascular adipose tissue in cardiovascular disease.
Recent Publications:
- A high-fat hypertensive diet induces a coordinated perturbation signature across cell types in thoracic perivascular adipose tissue, Terrian et al, 2026
- Phenotypic Changes in T Cell and Macrophage Subtypes in Perivascular Adipose Tissues Precede High-Fat Diet-Induced Hypertension, Kumar et al, 2021.
- Blood pressure changes PVAT function and transcriptome: use of the mid-thoracic aorta coarcted rat, Contreras et al, 2020.
Prediction of DNA – transcription factor binding
Predicting genome-wide transcription factor binding sites on DNA is a challenging problem. We are using machine learning tools to address this problem for members of the bHLH-PAS family of transcription factors.
Publications:
1. Prediction of mammalian tissue-specific CLOCK-BMAL1 binding to E-box motifs, Marri et al 2023.
2. Predictive Models of Genome-wide Aryl Hydrocarbon Receptor DNA Binding Reveal Tissue Specific Binding Determinants, Filipovic et al 2023.