A new tool published in Cell (2026) does exactly that — and you don’t need a chemistry degree to understand how it works.
The Problem: Drug Discovery is Painfully Slow
Finding a drug for a disease traditionally means:
- 🔬 Screening millions of compounds in the lab
- ⏳ Waiting years to see results
- 💰 Spending billions of rupees on R&D
And still — 90% of drug candidates fail.
The numbers paint a stark picture:
| Metric | Value |
|---|---|
| Years from lab to patient | 12+ |
| Average cost to develop one drug | ₹8,000 Cr |
| Drug candidates that fail | 90% |
So what if we could fundamentally change this equation?
Meet GPS: Gene Expression Profile Predictor on Chemical Structures
GPS is a new AI tool that connects the biology of disease directly to the chemistry of potential cures. The acronym breaks down as:
- G — Gene expression profile: How active each gene is in a diseased cell
- P — Predictor: An AI model that predicts what will happen
- S — Chemical Structures: The molecular shape of a drug candidate
In simple terms, GPS reads the “error log” of a diseased cell and then searches for the molecule that can undo the damage.
How GPS Works — In 3 Simple Steps
Step 1 — Read the Disease 🧬
GPS analyzes the gene expression pattern of diseased cells — basically reading the cell’s “error log” to understand what’s gone wrong at the molecular level.
Step 2 — Screen Millions of Molecules 🔍
It then scans a massive library of chemical structures and predicts which one will reverse the diseased gene pattern — like an “undo” button for the cell.
Step 3 — Rank the Best Candidates 💊
Using Monte Carlo tree search (a technique borrowed from AI chess!), GPS ranks and optimizes the top candidates for real-world drug potential.
Real Results: Liver Cancer (Hepatocellular Carcinoma)
HCC is the most aggressive form of liver cancer, with very limited treatment options today. Here’s what GPS achieved:
- 🔬 Novel Compounds Found: GPS identified new chemical series that no human researcher had specifically screened for HCC before.
- 🐭 Validated in Mice: The top candidates successfully reduced tumor size in preclinical mouse models.
- 🎯 Selective Activity: The compounds showed favorable selectivity — hitting cancer cells while sparing healthy tissue.
- ⚡ Speed of Discovery: What traditionally takes years was reduced to a computational screen in days.
Real Results: Lung Fibrosis (Idiopathic Pulmonary Fibrosis)
IPF causes irreversible lung scarring. No cure exists. Millions of patients worldwide are affected.
- 🔄 Repurposing Candidate Identified: GPS found an existing approved drug that could be repurposed for IPF — dramatically cutting development time.
- 🧫 New Anti-Fibrotic Molecule: A newly designed molecule lowered fibrosis markers in human lung tissue models — not just mice.
- 📊 Validated on Human Data: This is a critical step — validating AI predictions on human models before clinical trials.
Why This Matters for Pharma Graduates in India
GPS wasn’t built by computer scientists alone. It required deep expertise in areas that life science graduates already understand:
- Transcriptomics / Gene Expression — Understanding how genes behave in disease. This is core Pharma/Life Science knowledge.
- Cheminformatics — Representing chemical structures as data that computers can learn from.
- Deep Learning — The AI that learned patterns from millions of gene expression experiments.
- Virtual Screening — Using computers to screen millions of candidates before ever touching a test tube.
These are exactly the skills that Moleculytics teaches. The gap between traditional pharma knowledge and computational drug discovery is shrinking fast — and you can be part of it.
GPS is Fully Open Access — Use It Today, For Free
The best part? This isn’t locked behind a corporate paywall:
- 🌐 Web Portal: Run virtual screens directly in your browser. No installation needed.
- 💻 GitHub Code: Full source code released. Open for researchers and students.
- 📄 Published in Cell: Peer-reviewed. Citable. Real science — not just hype.
The Future of Pharma is Here
AI doesn’t replace the Pharma graduate’s knowledge of disease. It amplifies it.
GPS works because researchers understood what gene dysregulation means in cancer. The AI provided the speed and scale — but the scientific insight came from trained life scientists.
Which disease would you want GPS to screen for first? Share your thoughts with us on LinkedIn or reach out at hello@moleculytics.in.
Stay tuned to the Moleculytics blog for more breakdowns of cutting-edge AI × Pharma research, practical tutorials, and career insights for life scientists in India.