For research labsIndia-groundedCo-authorship · No IP grab

Collaborate with PetriDish

PetriDish is built for Indian biomedicine — 4.4M knowledge-graph edges, 55 PGx variants, 100 phytochemicals, 50 curated epifactors, all India-frequency-weighted. We're actively looking for senior PIs, hospital cohorts, and AYUSH-integration labs to co-author the validation work. Three open collaboration tracks below; pick the one that fits your lab and ping us.

Track 0 · TB Differential safety validation10–14 weeks to validation report

Sign-off PetriDish's safety-critical TB gate

PetriDish ships the only Indian autoimmune CDSS that refuses to give a risk score until TB is excluded (tb_diff.md v1.0). Logic is implemented; reviewer sign-off and a retrospective validation cohort are gating launch.

We bring
  • Full scoring engine — constitutional / respiratory / joint / epidemiological / lymph / lab
  • Red and amber flag overrides + state TB burden table (India TB Report 2023)
  • Output schema with clinician note + bilingual patient message + recommended investigations
  • Engineering bandwidth to recalibrate weights against your cohort
  • Co-author credit · ethics-committee paperwork drafting support
You bring
  • Senior rheumatologist OR senior infectious-disease physician sign-off (AIIMS / PGIMER / CMC equivalent)
  • Retrospective cohort: ≥ 50 confirmed TB arthritis · ≥ 30 Poncet's · ≥ 100 early RA · ≥ 20 TB+autoimmune · ≥ 50 normal controls
  • EMR / paper records access (de-identified)
  • Local ethics committee submission
Output: Validated module hitting ≥95% sensitivity for TB_FLAG, ≤5% false negative rate. Joint preprint in Indian J Rheumatol or Lancet South-East Asia.
Track 1 · Clinical validation8–12 weeks to preprint

Co-author a validation paper

PetriDish has 9 working modules, no clinical validation paper yet. We'd like to fix that with your patient cohort + your senior author position.

We bring
  • Knowledge graph + 55 PGx variants + IndiGen frequencies
  • Live API endpoints (sub-200ms PGx safety check)
  • Engineering time to build the cohort-specific dashboard
  • Statistical analysis + preprint drafting support
  • Co-author credit · no IP claim on your cohort data
You bring
  • Retrospective or prospective cohort (>100 patients ideal)
  • Pharmacy / lab data linked to outcomes
  • Ethics committee clearance for retrospective use
  • Senior-author position
Output: Joint preprint within 8–12 weeks; pursue a Lancet South-East Asia / Indian J Pharmacol submission.
Track 2 · Phytochemical curation4–6 weeks to first launch

Help us scale IMPPAT 100 → 500

PetriDish has 100 curated phytochemicals with SMILES + 192 MAMMAL DTI predictions. ACTREC, IIIT-D, and CDFD have larger Indian phytochemistry corpora. Cross-licence them in.

We bring
  • Existing IMPPAT 100 + MAMMAL DTI scoring infra
  • Pipeline to ingest CSV → Neo4j → API in <1 day
  • PubMed citation annotation per compound
  • Visible attribution: your lab as a data source on /methods
You bring
  • Public or shareable phytochemistry dataset
  • SMILES, herb-of-origin, optional bioactivity priors
  • Lab review of our drug-target predictions for your top 20 compounds
Output: Joint /ayurbridge product launch; PetriDish becomes the search layer over your data.
Track 3 · Multi-omics integration6–10 weeks to working demo

Bring scGPT / methylome / spatial

EpiOnco today is a curated knowledge layer over published Indian signatures. If your lab runs scRNA-seq, EPIC arrays, or spatial — there's a path to ingest cohort embeddings into the Tumour Ability Score.

We bring
  • Tumour Ability Score framework (India delta)
  • 14 cancer hallmarks + 50 epifactor knowledge layer
  • Engineering to wire your embeddings into TAS computation
  • Patient-facing TAS brief PDF generator
You bring
  • Sample-level embeddings or normalized expression / methylation
  • Cohort metadata (cancer type, stage, treatment, outcome)
  • Bioinformatician collaboration during integration
Output: Cohort-specific EpiOnco that the senior PI demos at AIIMS / Tata Memorial.
Track 5 · Ayurvedic–PGx interaction landscape16–24 weeks to atlas v1

Characterise India's herb-drug interaction blind spot

AutoCure carries 30+ Ayurvedic contraindications today (Ashwagandha in SLE, Curcumin + warfarin, Piperine + MTX via CYP3A4). They're sourced from PMIDs but the systematic landscape — top 20 Ayurvedic compounds × top 10 autoimmune drugs across CYP2D6/3A4/2C19 — has never been characterised. india_autoimmune.md §14.5 frames it as the open research question.

We bring
  • 12 curated compounds already in the knowledge graph (pathway + targets + PMIDs)
  • PGx variant overlay (61 variants, IndoPGx /pgx flagship)
  • Computational docking pipeline scaffold
  • Co-authorship + AYUSH ministry / NMPB grant drafting support
You bring
  • Wet-lab CYP inhibition assay capacity (CYP2D6 / 3A4 / 2C19 microsomes)
  • Optional: clinical pharmacology PI for healthy-volunteer PK study (curcumin + piperine + methotrexate)
  • Ethics committee + DCGI study approval
Output: First systematic Indian herb-PGx-drug interaction atlas. AYUSH ministry / NMPB grant submission + 2 first-author papers (in vitro + in silico).
Track 6 · Indian immunopeptidomics partnership6–12 months to v1 dataset

Generate the proprietary Indian immunopeptidome — the durable autoimmune moat

Mass-spec immunopeptidomics (eluted ligands from Indian HLA-typed cells) is THE durable moat for India autoimmune AI — it cannot be replicated without Indian samples and it improves every prediction layer (per autoimmune_protein_deep_research.md §10 Phase 4). PetriDish has the computational engine and Protein Atlas curated knowledge layer. We need a wet-lab partner.

We bring
  • Protein Atlas curated knowledge layer (autoantigens with PTM, mimicry maps, open wedges)
  • Pan-allele peptide-MHC presentation pipeline (MHCflurry / ESMCBA — commercial-clean)
  • Indian HLA frequency context (IndiGen + DATRI + published cohorts)
  • Computational follow-on: present-peptidome × autoantigen prediction × mimicry screen
  • Public dataset published as Cell / Nature Immunology / Nature Methods candidate
  • Co-authorship · no IP claim on samples
You bring
  • HLA immunoprecipitation + mass-spec from 100–300 Indian donors (top-5 frequency HLA-DR + HLA-B alleles)
  • Wet-lab capacity (immunology / proteomics core)
  • Senior PI position
  • Ethics committee + DCGI study approval
  • Sample collection logistics
Output: First public Indian immunopeptidome dataset. Cell / Nat Immunol / Nat Methods submission. DST-SERB + ICMR + private foundation grant fit. Permanent proprietary moat for downstream Indian autoimmune AI.
Track 7 · UK Biobank SAS subset + Indian autoimmune validation4-6 months to validation paper (assuming UKB approval lands)

Validate PetriDish's HLA-imputation × India-calibrated PRS engine — zero wet lab needed

The cleanest validation cohort for the IndiaAuto Risk Panel exists today, in public: the UK Biobank South Asian subset (~50,000 individuals with both genotype data and clinical phenotypes). Run the Phase 2 SNP2HLA + PRS-CSx pipeline on their genomes, compare predictions to documented RA / SLE / AS / T1D / IBD diagnoses. No clinical access. No wet lab. The unlock is a UK Biobank Researcher application + ~3 months compute.

We bring
  • Phase 1 curated risk-loci engine (live now)
  • Phase 2 SNP2HLA / CookHLA pipeline scaffold — South Asian reference panel build
  • Phase 2 PRS-CSx multi-ancestry calibration pipeline
  • Phase 3 GNN training infrastructure
  • Validation paper drafting + figure generation
  • Co-authorship · no IP claim on UKB data
You bring
  • UK Biobank Researcher account (~6-8 week application — we draft it)
  • Wet/dry-lab PI with sponsorship eligibility (IISc / NCBS / IIIT-D / TIFR)
  • Computational geneticist collaborator (PRS-CSx familiarity helpful)
  • Optional: Indian rheumatology cohort with HLA typing for the final validation step
Output: PetriDish's risk panel becomes the first validated India-calibrated autoimmune PRS published. Target: Nature Genetics / European J of Human Genetics / Genome Medicine. BIRAC + DBT grant fit. Converts the Phase 1 curated engine into a published, validated, defensible product.

Five live use cases

Concrete starting points if you're not sure which track fits. Each maps to a real PetriDish module already running in production.

🧬
Pediatric leukemia PGx audit
Retrospective NUDT15 / TPMT genotype vs maintenance-phase ANC trajectory in 200 B-ALL kids. We do the PGx analysis, you publish.
Fit · Pediatric oncology · Hematology
💊
Stent thrombosis CYP2C19 cohort
Post-PCI patients on clopidogrel. Match CYP2C19*2/*3 status to MACE outcomes. India under-represented in trial-derived guidelines.
Fit · Interventional cardiology
🌿
Ayurvedic herb-CYP interaction registry
Build the first Indian registry of clinically documented herb-drug interactions. Start with the 12 we&apos;ve curated. Grow to 100 over 6 months.
Fit · Pharmacology · AYUSH integration
🔬
Indian OSCC methylome panel
Validate the hypomethylated p53 / DNMT3A signature in a fresh cohort. PetriDish provides the EpiOnco scoring layer.
Fit · Head-neck oncology · Epigenomics
🏥
Hospital PGx CDSS pilot
Embed /pgx/check into your EHR prescribing flow. Measure alert acceptance, prescription changes, and patient outcomes over 6 months.
Fit · Hospital informatics · CMIO

Get in touch

Email Shailesh with your lab, cohort size, and which track interests you. Reply within 48 hours. If there's a fit, we'll set up a 30-min Zoom + share a one-page MoU template.