Bioinformatics & AI-Driven Drug Discovery
This project applies AI, bioinformatics, and computational chemistry to accelerate how we discover, prioritise, and design new drugs—moving from raw omics data to target discovery, hit identification, and lead optimisation pipelines.
Project Overview
The Bioinformatics & AI-Driven Drug Discovery initiative focuses on building modular tools and pipelines that can process biological data (genes, proteins, pathways) and map it to potential therapeutic targets and candidate molecules.
Current focus tracks:
- Target discovery using omics data, disease networks, and pathway analysis.
- Virtual screening & scoring of compounds against targets using AI-assisted docking and QSAR models.
- Lead optimisation workflows that propose and rank molecular modifications for better potency and ADMET.
Objectives
- Reduce time and cost of early-stage discovery using in-silico and AI-first approaches.
- Provide transparent, reproducible pipelines for target and molecule prioritisation.
- Integrate multiple data types (omics, structures, literature) into unified views.
- Build tools that can support both academic labs and industry partners.
Tech Stack & Methods
The project combines bioinformatics, cheminformatics, and machine learning:
- Data: Genomic and transcriptomic profiles, protein structures, pathway databases, chemical libraries, clinical and literature-derived evidence.
- Bioinformatics: Differential expression, network analysis, pathway enrichment, and target prioritisation scores.
- Cheminformatics & modelling: Molecular descriptors & fingerprints, docking, molecular dynamics (where feasible), and generative models for molecules.
- ML models: QSAR/MLP/GNN models for activity & ADMET, ranking models for candidate selection, and uncertainty-aware predictions.
Example Use Cases
- Identifying new targets or repositioning opportunities for existing drugs based on omics data.
- Virtual screening of large libraries to find top candidate hits for wet-lab follow-up.
- Ranking & optimising analogs for potency, selectivity, and drug-likeness before synthesis.
- Building disease-specific knowledge graphs linking genes, pathways, compounds, and phenotypes.
Safety, Ethics & Validation
The project emphasises scientific rigour and ethics: models are used to prioritise hypotheses, not to claim clinical efficacy on their own. Wet-lab validation, expert review, and regulatory considerations remain central in any downstream deployment.
Social Plugin