Bioinformatics Seminars

Bioinformatics Seminar

Time: 10:45am Tuesdays.
Level 7 Seminar Room 2, WEHI1

20 March 2018

Inferring edge function in protein-protein interaction networks

Daniel Esposito
WEHI Bioinformatics

Post-translational modifications (PTMs) regulate many key cellular processes. There has been increasing attention in using information rich databases such as the Gene Ontology (GO) in protein-protein interaction (PPI) and protein function prediction tasks. However, these methods fail to give insight in the biochemical nature of these interactions, and more so fail to give insight regarding the dynamic nature of protein function in the context of an interaction. We present a machine learning approach that uses information from Gene Ontology (GO), InterPro and Pfam protein annotations to infer edge functions in protein-protein interaction networks. In our approach we train a One-Vs-All estimator on multi-labelled PPIs mined from the Kyoto Encyclopedia Of Genes and Genomes (KEGG) and Human Protein Reference Database (HPRD), achieving a 5-fold cross-validation F1 score of 0.86. Furthermore, when applied to the human interactome our method confidently classifies 62% of interactions at a probability of 0.7 or higher.

Search past seminars