PertInInt: An Integrative, Analytical Approach to Rapidly Uncover Cancer Driver Genes with Perturbed Interactions and Functionalities

Abstract

SUMMARY A major challenge in cancer genomics is to identify genes with functional roles in cancer and uncover their mechanisms of action. We introduce an integrative framework that identifies cancer-relevant genes by pinpointing those whose interaction or other functional sites are enriched in somatic mutations across tumors. We derive analytical calculations that enable us to avoid time-prohibitive permutation-based significance tests, making it computationally feasible to simultaneously consider multiple measures of protein site functionality. Our accompanying software, PertInInt, combines knowledge about sites participating in interactions with DNA, RNA, peptides, ions, or small molecules with domain, evolutionary conservation, and gene-level mutation data. When applied to 10,037 tumor samples, PertInInt uncovers both known and newly predicted cancer genes, while additionally revealing what types of interactions or other functionalities are disrupted. PertInInt’s analysis demonstrates that somatic mutations are frequently enriched in interaction sites and domains and implicates interaction perturbation as a pervasive cancer-driving event.

Learning probabilistic protein-DNA recognition codes from DNA-binding specificities using structural mappings

“…While de novo DNA-binding specificity prediction is necessary for predicting binding preferences of completely uncharacterized TFs, in the context of natural variation and disease, there is great interest in predicting the changes in DNA-binding specificities induced via single-amino acid alterations to wildtype TFs. Mutated TFs are seen in cancer (Kobren et al 2020), and in inherited diseases (Hamosh et al 2005;Chi 2006). If such an alteration occurs in a DNA-contacting residue, as a first approximation, de novo prediction is necessary only for the binding site positions contacted by that altered residue, while the specificity for the remaining binding site positions can be transferred directly from the wildtype.…”

“…While de novo DNA-binding specificity prediction is necessary for predicting binding preferences of completely uncharacterized TFs, in the context of natural variation and disease, there is great interest in predicting the changes in DNA-binding specificities induced via single-amino acid alterations to wildtype TFs. Mutated TFs are seen in cancer (Kobren et al . 2020), and in inherited diseases (Hamosh et al .…”

“…However, despite these extensive catalogues of DNA-binding specificities for wildtype TFs, as yet very few methods have been able to use these data in order to uncover the underlying determinants of protein-DNA specificity; this prevents both the prediction of de novo specificities for uncharacterized TFs (e.g., including those in non-model organisms or those that have proved difficult to study experimentally) as well as the prediction of altered DNA-binding specificities due to mutations or SNPs within TFs (e.g., as observed in several Mendelian disorders (Veraksa et al . 2000), cancers (Kobren et al . 2020), and across healthy populations (Barrera et al .…”

dSPRINT: predicting DNA, RNA, ion, peptide and small molecule interaction sites within protein domains

“…While we have demonstrated the usefulness of our predictions for annotating DUF domains, our framework can be used for annotating many types of other protein domains, including assigning novel functions for already annotated domains. Moreover, knowledge of the specific sites within domains involved in interactions can help prioritize disease-causing mutations ( 16 , 62 ) and may result in identifying novel drug targets, as most drugs operate by competing for ligand-binding sites ( 63 ).…”

“…Recently, it has been shown that domains provide a framework within which to aggregate structural co-complex data and this per-domain aggregation allows accurate inference of positions within domains that participate in interactions ( 15 ). Domains annotated in this manner can then be used to identify interaction sites within proteins, and such domain-based annotation of interaction sites has been the basis of approaches to detect genes with perturbed functionalities in cancer ( 16 ) and to perform systematic cross-genomic analysis of regulatory network variation ( 17 ).…”