Newfound coding potential of transcripts unveils missing members of human protein communities

Recent proteogenomic approaches have led to the discovery that regions of the transcriptome previously annotated as non-coding regions (i.e. UTRs, open reading frames overlapping annotated coding sequences in a different reading frame, and non-coding RNAs) frequently encode proteins (termed alternative proteins). This suggests that previously identified protein communities are partially incomplete since alternative proteins are not present in conventional protein databases. Here we incorporate this increased diversity in the re-analysis of a high throughput human network proteomics dataset thereby revealing the presence of 203 alternative proteins within 163 distinct communities associated with a wide variety of cellular functions and pathologies. We found 19 genes encoding both an annotated (reference) and an alternative protein interacting with each other. Of the 136 alternative proteins encoded by pseudogenes, 38 are direct interactors of reference proteins encoded by their respective parental gene. Finally, we experimentally validate several interactions involving alternative proteins. These data improve the blueprints of the human protein-protein interaction network and suggest functional roles for hundreds of alternative proteins.