Publication

Journal	Briefings in bioinformatics
Page(s)	1
Title of proceedings	Briefings in bioinformatics
DOI	10.1093/bib/bbx029

URL	https://academic.oup.com/bib/article/19/5/995/3091148
Type of Open Access	Green OA Embargo (Freely available via Repository after an embargo)

Various techniques have been developed for identifying the most probable interactants of a protein under a given biological context. In this article, we dissect the effects of the choice of the protein-protein interaction network (PPI) and the manipulation of PPI settings on the network neighborhood of the influenza A virus (IAV) network, as well as hits in genome-wide small interfering RNA screen results for IAV host factors. We investigate the potential of context filtering, which uses text mining evidence linked to PPI edges, as a complement to the edge confidence scores typically provided in PPIs for filtering, for obtaining more biologically relevant network neighborhoods. Here, we estimate the maximum performance of context filtering to isolate a Kyoto Encyclopedia of Genes and Genomes (KEGG) network Ki from a union of KEGG networks and its network neighborhood. The work gives insights on the use of human PPIs in network neighborhood approaches for functional inference.