PAIR

PAIR provides a comprehensive dataset of experimentally reported and predicted protein–protein interactions in Arabidopsis thaliana (5,990 experimentally validated interactions and 145,494 predicted interactions) to support interactome analysis.

Key Features:

• Experimental and predicted interactions: Integrates 5,990 experimentally validated interactions with 145,494 predicted protein–protein interactions for Arabidopsis thaliana.
• Prediction methodology: Uses a fine-tuned support vector machine (SVM) model that integrates gene co-expressions, domain interactions, shared Gene Ontology (GO) annotations, co-localizations, phylogenetic profile similarities, and homologous interactions in other organisms (interologs).
• Coverage and reliability: Predicts approximately 24% of the Arabidopsis interactome with an estimated reliability of 44%, supported by validation using two independent example datasets.

Scientific Applications:

• Gene function inference: Provides interaction evidence to support the functional annotation of Arabidopsis genes.
• Regulatory network elucidation: Enables reconstruction and analysis of molecular regulatory networks in Arabidopsis thaliana.
• Target identification: Supports identification of candidate targets for genetic manipulation or chemical intervention.
• Hypothesis generation: Facilitates generation of hypotheses about unknown protein functions and novel protein–protein interactions.

Methodology:

Predictions are produced by a fine-tuned support vector machine (SVM) integrating gene co-expression, domain interactions, shared GO annotations, co-localizations, phylogenetic profile similarities, and interologs, with validation performed using two independent example datasets.