A comprehensive perturbation-signature search engine integrating connectivity maps from L1000, Tahoe-100M, NIBR DRUG-seq, CMap, Ginkgo Bioworks, SciPlex, DeepCover MoA, Perturb Atlas, Replogle Perturb-seq, CREEDS, and RummaGEO.
- Production: https://perturbseqr.maayanlab.cloud
- Dev: https://perturbseqr.k8s.dev.maayanlab.cloud
- GraphQL API: https://perturbseqr.maayanlab.cloud/graphiql
Runs the stack via Docker Compose, using AWS S3 for ingest, downloads, and enrichment indexes. The dev cluster retains MinIO; see DEV_DEPLOY.md.
cp .env.example .env
# Edit .env: supply AWS credentials and set ENRICH_S3_PREFIX to a unique
# enrich/local-<identifier>/ prefix for this database (never enrich/production/).
docker compose build perturbseqr-enrich ingest
docker compose up -d
# Read GMTs already stored at the root of the AWS perturbseqr bucket
COMPOSE_PROFILES=jobs docker compose run --rm ingest- App: http://localhost:3000
- GraphiQL: http://localhost:3000/graphiql
Run only the backend services in docker; run the Next.js app with hot reload:
cp .env.example .env
# Supply AWS credentials and a unique local ENRICH_S3_PREFIX as above
docker compose build perturbseqr-enrich
docker compose up -d perturbseqr-postgres perturbseqr-enrich
npm install && npm run devWipe local PostgreSQL volumes and rebuild from the existing AWS GMTs. This leaves AWS objects intact:
docker compose down -v
docker compose up -d
COMPOSE_PROFILES=jobs docker compose run --rm ingestSee DEV_DEPLOY.md for the dev Kubernetes cluster deployment workflow.
See mcp/README.md for setup and supported tools, and SKILL.md for the Perturb-Seqr agent skill.
Perturb-Seqr: Comprehensive Signature Search Engine Integrating Connectivity Maps from Multiple Sources John K. Gardner, Lily D. Taub, Daniel J. B. Clarke, Ido Diamant, Avi Ma'ayan Department of Pharmacological Sciences, Mount Sinai Center for Bioinformatics, Icahn School of Medicine at Mount Sinai, New York, New York; CFDE Data Resource Center