STITCH

STITCH maps known and predicted interactions between chemicals, including drugs, and proteins to enable integrated analysis of molecular interaction networks.

Key Features:

• Extensive interaction network: Contains over 390,000 chemicals and 3.6 million proteins from 1,133 organisms, with 367,000 high-confidence human protein–chemical interactions (a 45% increase reported vs. previous versions).
• Integration of experimental and curated evidence: Aggregates interaction evidence from metabolic pathways, crystal structures, binding experiments, drug–target relationships, and phenotypic effects.
• Database and literature sources: Incorporates data from BindingDB, PharmGKB, the Comparative Toxicogenomics Database, STRING interactions, and text mining of full-text articles.
• Chemical similarity and structure-based predictions: Uses chemical structure similarities to support prediction of interactions and to augment text-mining results.
• Orthology-based transfer: Transfers interactions between species using an orthology-focused approach that emphasizes orthologous proteins over paralogous ones.
• Stereochemistry handling: Provides representation options that merge or separate stereoisomers, with explicit consideration for carbohydrates and chiral drugs.
• Evidence provenance: Links each interaction to its original data sources for traceability of supporting evidence.

Scientific Applications:

• Pharmacology: Analysis of drug–target relationships and prediction of putative drug interactions with proteins.
• Toxicogenomics: Mapping chemical–phenotype associations and assessing chemical effects across organisms.
• Molecular biology and network analysis: Large-scale reconstruction and interrogation of protein–chemical interaction networks across species.
• Comparative and cross-species studies: Use of orthology-based transfers to compare interactions within protein families across organisms.

Methodology:

Integrates interaction data from metabolic pathways, crystal structures, binding experiments, drug–target relationships, and phenotypic effects; aggregates databases including BindingDB, PharmGKB, the Comparative Toxicogenomics Database and STRING; applies text mining of full-text articles and chemical structure-similarity analyses; transfers interactions across species using an orthology-based method and supports merging or separating stereoisomers.