shrut27's picture
Implement complete API contract, persistence, jobs, artifacts, and architecture status
ff453a6 verified
|
Raw
History Blame Contribute Delete
2.01 kB
---
title: TP53 Mutant Discovery Platform
emoji: 🧬
colorFrom: blue
colorTo: green
sdk: static
pinned: false
---
# AI-Guided TP53 Mutant Discovery and Drug Repurposing Platform
Interactive research dashboard for mutation-specific TP53 variant routing, wild-type versus mutant structural comparison, AI virtual screening, matched docking, molecular-dynamics triage, ADMET filters, and consensus Rescue / Opportunity ranking.
## What it covers
- Variant intelligence and QC (HGVS normalization, type mismatches, domain mapping)
- Mutation-specific routing (do not dock truncating / splice variants by default)
- WT vs mutant structure and RMSF comparison
- Pocket / druggability gates
- Compound library (approved, clinical, screening, Y220C controls)
- AI prescreen with confidence and applicability
- Matched WT vs mutant docking and interaction fingerprints
- Replica MD, contact occupancy, MM/GBSA proxies
- Multi-objective Rescue / Opportunity Score with adjustable weights
- Validation metrics, risk register, service/API map
- Per-variant PDF report and CSV exports
## REST API
The Docker runtime serves the dashboard and a versioned FastAPI backend from the
same container. OpenAPI documentation is available at `/docs` and `/openapi.json`.
Implemented contracts:
- `POST /projects`
- `POST /variants:ingest`
- `POST /variants/{id}/analyze`
- `POST /structures:prepare`
- `POST /pockets:detect`
- `POST /compounds:screen`
- `POST /docking:run`
- `POST /md:run`
- `POST /rankings:compute`
- `GET /jobs/{id}` and `GET /projects/{id}/jobs`
- `GET /projects/{id}/artifacts` and `GET /artifacts/{id}`
- `GET /projects/{id}/report`
Long-running contracts return `202` plus a `job_id`. SQLite stores project
records, job status/logs, versioned outputs, and downloadable JSON artifacts.
Run locally:
```bash
uvicorn api.main:app --host 0.0.0.0 --port 7860
```
## Notes
All scores are computational estimates for prioritization. They are not proof of binding, p53 rescue, or clinical efficacy.