Research Projects
Accelerating drug discovery through computational approaches, AI, and bioinformatics
AI-Based ADME-Tox Prediction System
Built enterprise-grade toxicity prediction system analyzing 1,067 molecules across 7 endpoints (hERG, solubility, mutagenicity, carcinogenicity, hepatotoxicity). Achieved 90.7% average accuracy with Random Forest models and R² = 0.989 for solubility prediction.
Technologies Used:
Key Highlights:
- 90.7% avg accuracy
- 7 ADME-Tox endpoints
- 1,067 molecules dataset
- 15 publication-quality visualizations
AI Classifier for Rare Cancer Diseases
Developing deep learning classifier for rare cancer detection using DNA methylation data. Combining epigenomics with machine learning for early cancer diagnosis and precision medicine applications.
Technologies Used:
Key Highlights:
- MS research project
- DNA methylation analysis
- Rare cancer detection
- Precision medicine
QSAR Modeling for Drug Discovery
Developed predictive QSAR models using Mordred descriptors and machine learning to predict IC50 values for potential drug candidates. Includes analog generation, SHAP interpretation, and lead optimization.
Technologies Used:
Key Highlights:
- 95% prediction accuracy
- SHAP feature importance
- Analog optimization
Molecular Docking & Virtual Screening
Conducted structure-based drug design using PyRx, GOLD, and AutoDock Vina. Performed consensus scoring, RMSD clustering, and PLIF analysis for DAT inhibitors and other therapeutic targets.
Technologies Used:
Key Highlights:
- Consensus scoring
- RMSD clustering
- 2D/3D interaction mapping
Transcriptomics & Gene Expression Analysis
Analyzed GEO datasets (e.g., GSE183795 pancreatic cancer) for expression normalization, metadata curation, and ML-based disease classification with interpretable results.
Technologies Used:
Key Highlights:
- Expression profiling
- Disease classification
- Biomarker identification
Metagenomics & Microbiome Analysis
Conducted comprehensive microbiome data analysis, taxonomic classification, diversity metrics, and functional profiling for understanding microbial communities and their impact on health.
Technologies Used:
Key Highlights:
- Taxonomic profiling
- Alpha/Beta diversity
- Functional annotation
Research Philosophy
"I bridge the gap between biology and artificial intelligence — transforming complex biological data into actionable insights through computational drug design, QSAR modeling, and AI-driven toxicity prediction. From DNA methylation patterns in rare cancers to ADME-Tox profiling of drug candidates, my work combines scientific rigor with explainable AI (SHAP, attention mechanisms), enabling researchers to accelerate drug discovery with transparency and confidence. I turn molecules into medicine, faster."