Evidence provenance

Data Sources & Methods

Every genomics and pharmacogenomics tool in this platform discloses exactly what's a live external data call versus a computational heuristic. This page lists every live integration in one place, with what it's used for and where the underlying data comes from.

Live external data sources

All queried live at request time (with short-lived caching to respect rate limits), not pre-downloaded snapshots.

Open Targets Platform

Target-disease genetic-association evidence, tractability (small molecule / antibody / PROTAC), known drugs and clinical candidates, safety liabilities.

Used in: Genetic Validation, Target Assessment, Virtual Perturbation Lab, Lead Optimization (target-context scoring)

gnomAD

Gene-level population constraint (pLI, missense z-score, observed/expected ratios) and variant population allele frequencies, including South Asian.

Used in: Genetic Validation, Target Assessment, Variant Annotation

ClinVar (via NCBI E-utilities)

Known clinical-significance-classified variant records for a gene.

Used in: Genetic Validation, Target Assessment, Variant Annotation

GWAS Catalog (EBI)

Published genome-wide association study hits for a gene.

Used in: Genetic Validation, Target Assessment

GenomeIndia / GI-DB (IGIB)

India's national genomic database -- 130M+ variants from 9,768 individuals across 83 ethnolinguistic groups. Queried for the specific rsIDs defining the star alleles used in dosing guidance.

Used in: Pharmacogenomics / Precision Dosing (all 13 gene-drug pairs)

PharmGKB

Clinical annotation counts and evidence levels linking a gene to drug-response phenotypes -- the database CPIC guidelines are published through.

Used in: Pharmacogenomics / Precision Dosing

Ensembl VEP

Variant consequence annotation (gene, impact, consequence type), embedding ClinVar significance and gnomAD population frequency in a single lookup.

Used in: Variant Annotation

Enrichr

Gene set over-representation analysis against GO Biological Process, KEGG, and Reactome.

Used in: Toxicity Command Center (transcriptome-wide enrichment)

AlphaFold DB

Predicted protein structures (pLDDT confidence) for pocket detection and MD/FEP readiness triage.

Used in: structure-based docking/pocket-detection workflows

Real computational methods (no external API, but genuine statistics/algorithms)

Differential expressionMedian-of-ratios normalization + Welch's t-test + Benjamini-Hochberg FDR correction on user-supplied replicate count data.
Ranked GSEAgseapy prerank (Kolmogorov-Smirnov-based enrichment score, permutation-tested) on the full ranked gene list.
PCAz-score-standardized principal component analysis for expression-data QC (sample clustering, outlier detection).
Benchmark dose (BMD/BMDL)Four-parameter Hill-model nonlinear regression with bootstrap-resampled lower bound.
PGx dosing logicPublished CPIC guideline recommendations by gene-drug pair, not a locally-fit model.

Pharmacogenomic gene-drug pairs (CPIC-guideline)

13 pairs, each citing its specific CPIC guideline publication -- see the Pharmacogenomics / Precision Dosing tool for full guidance text and citations per pair.

Tacrolimus / CYP3A5 · Carbamazepine / HLA-B*15:02 · Phenytoin / CYP2C9 · Warfarin / CYP2C9+VKORC1 · Clopidogrel / CYP2C19 · Voriconazole / CYP2C19 · Codeine / CYP2D6 · Atomoxetine / CYP2D6 · Abacavir / HLA-B*57:01 · Allopurinol / HLA-B*58:01 · Azathioprine-Mercaptopurine / TPMT+NUDT15 · Fluoropyrimidines / DPYD · Simvastatin / SLCO1B1

Decision-support only. Every tool on this page is intended for research, target/candidate prioritization, and clinical development planning -- not autonomous prescribing or diagnosis. Population-prior estimates are literature-derived starting points, not diagnoses. Verify against current CPIC/PharmGKB guidance and use qualified clinical/scientific judgment before acting on any output.