Molecular grammars of intrinsically disordered regions that span the human proteome
Intrinsically disordered regions (IDRs) of proteins are defined by functionally relevant molecular grammars. This refers to IDR-specific non-random amino acid compositions and non-random patterning of distinct pairs of amino acid types. Here, we introduce GIN (Grammars Inferred using NARDINI+) as a resource, which we have used to extract the molecular grammars of all human IDRs and classified them into thirty distinct clusters. Unbiased analyses of IDRome-spanning grammars reveals that specialized IDR grammar features direct biological processes, cellular localization preferences, and molecular functions. IDRs with exceptional grammars, defined as sequences with high-scoring non-random features, are harbored in proteins and complexes that enable spatial and temporal sorting of biochemical activities. Protein complexes within the nucleus recruit specific factors through top-scoring IDRs. These IDRs are frequently disrupted via cancer-associated mutations and fusion oncoproteins. Overall, GIN enables the decoding of sequence-function relationships of IDRs and can be deployed in IDR-specific and IDRome-wide analyses.