Deep Learning-Driven AMR Drug Repurposing
A deep learning framework to identify FDA-approved non-antibiotic drugs with potential antibacterial activity against ESKAPE pathogens and M. tuberculosis. Leveraging ChEMBL bioactivity data and molecular fingerprints, the project uses an ensemble of Random Forest and Deep MLP classifiers with ChemBERTa integration to predict drug efficacy and interpret molecular features via SHAP.
TAM-Atlas: Tumor-Associated Macrophage Heterogeneity Across Cancers
A computational framework analyzing tumor-associated macrophage (TAM) heterogeneity across 8 cancer types to predict immunotherapy response using single-cell transcriptomics. Integrates ~640,000 cells from published datasets through a seven-phase pipeline, identifying 6 conserved macrophage states and linking SPP1+ and TREM2+ populations to checkpoint inhibitor resistance. Employs scVI deep generative modeling, gene regulatory network inference, and Open Targets druggability prioritization.
Single-Cell Atlas of COVID-19 Lung Pathology
Reanalysis of single-nucleus RNA-seq data from Melms et al. (2021) examining lung tissue from 26 COVID-19 patients and healthy controls. Identifies myeloid expansion, epithelial depletion, and pro-fibrotic remodeling across 94,027 cells using scVI batch correction and Leiden clustering, with an interactive 8-tab Shiny app for exploring gene expression and cell-cell interactions.
DeepAMR: AI-Powered Antimicrobial Resistance Prediction
A deep learning platform that analyzes genomic sequences to predict antimicrobial resistance patterns, built for the Bangladesh healthcare system. Delivers AMR predictions in under 15 minutes with 95%+ accuracy, identifying resistance genes and mutations across 12+ organisms and 50+ antibiotics — reducing traditional culture-based testing from days to hours.
Fezf2-Mediated Cortical Development: Multi-Omics Analysis
Integrative analysis of bulk and single-cell RNA-Seq data to elucidate Fezf2’s role in cortical neurogenesis. Leveraging the GSE153164 dataset (Di Bella et al. 2021) to investigate temporal and cell-type-specific mechanisms by which Fezf2 mutations disrupt cortical development.
Single-Cell Meta-Analysis of Microglial Activation in Alzheimer’s Disease
Meta-analysis of single-cell RNA-Seq datasets to identify microglial states and activation patterns in Alzheimer’s disease. Integrating multiple cohorts to characterize disease-associated microglia subtypes and their transcriptomic signatures across disease stages.
Integrative Spatial-scRNA-seq Atlas of Immunotherapy Resistance Across Cancer Types
Comprehensive spatial transcriptomics and single-cell analysis to map tumor microenvironment and resistance mechanisms in immunotherapy. Investigating how cell-type interactions influence tumor development and immunotherapeutic response, including SPP1-expressing macrophage-mediated resistance through adenosine pathway activation.