Inferring Protein Domain Semantic Roles Using word2vec

Abstract

In this paper, using word2vec, we demonstrate that proteins domains may have semantic \"meaning\" in the context of multi-domain proteins. Word2vec is a group of models which can be used to produce semantically meaningful embeddings of words or tokens in a vector space. In this work we treat multi-domain proteins as \"sentences\" where domain identifiers are tokens which may be considered as \"words\". Using all Interpro (Finn, Attwood et al. 2017) eukaryotic proteins as a corpus of \"sentences\" we demonstrate that Word2vec creates functionally meaningful embeddings of protein domains. We additionally show how this can be applied to identifying the putative functional roles for Pfam (Finn, Coggill et al. 2016) Domains of Unknown Function.