KUPS

KUPS aggregates protein–protein interaction data and constructs datasets of interacting protein pairs (IPPs) and non-interacting protein pairs (NIPs) with associated features to support development and evaluation of computational protein–protein interaction (PPI) prediction models.

Key Features:

• Extensive Database Integration: Aggregates data from IntAct, HPRD, MINT, UniProt, and Gene Ontology, comprising 185,446 IPPs and approximately 1.5 billion NIPs.
• Customizable Data Sets: Allows filtering by model organism, interaction type, and experimental evidence, and offers four distinct strategies to generate NIPs to mitigate biased estimation in PPI prediction.
• Feature Provision: Extracts and provides features associated with both IPPs and NIPs for use in computational modeling.
• Benchmark Data Sets: Provides two benchmark data sets for comparative algorithm evaluation.

Scientific Applications:

• Model Development: Enables training and validation of computational models for predicting protein–protein interactions using curated IPP and NIP datasets.
• Algorithm Evaluation: Supports comparative evaluation of PPI prediction algorithms using provided benchmark data sets.
• Biological Research: Facilitates studies of cellular processes, disease mechanisms, and identification of potential therapeutic targets through curated interaction data.

Methodology:

KUPS integrates records from IntAct, HPRD, MINT, UniProt, and Gene Ontology; constructs IPP and NIP datasets; applies filters for model organisms, interaction types, and experimental evidence; implements four NIP-generation strategies; extracts associated features; and provides two benchmark data sets.