Probabilistic approaches to inference of mutation rate and selection in cancer

Organizer: IRB Barcelona

Date: Friday 18 June 2021, 12.00h

Title:"Probabilistic approaches to inference of mutation rate and selection in cancer"

Host: Núria López-Bigas, PhD. Biomedical Genomics Lab -  Cancer Science Programme - IRB Barcelona

Biomed Webinar

Abstract:

Cancer is a highly complex system that evolves asexually under high mutation rates and strong selective pressures. Cancer genomics efforts have identified genes and regulatory elements driving cancer development and neoplastic progression. The detection of both significantly mutated (positive selection) and undermutated (negative selection) genes is completely confounded by the genomic heterogeneity of the cancer mutation rate. Here, I present an approach to addressing mutation rate heterogeneity to increase the power and accuracy of selection inference. Using a hierarchical model, we infer the distribution of mutation rates across genes that underlies the observed distribution of the synonymous mutation count within a given cancer type. This enables the inference of the probability of nonsynonymous mutations under neutrality without additional parameters, however explicitly taking into account cancer-type-specific mutational signatures, which are known to be highly distinct. In addition to detecting an excess in the total number of mutations, we then augmented our test through integrating information at the intra-gene level. Based on a mutational model that accounts for the extended sequence context (>5-mers) around mutated sites, this second component of the test identifies genes with an excess of mutations in unusual nucleotide contexts, which deviate from the characteristic context around neutrally evolving passenger mutations. I will show that the inclusion of this context test increases power to detect cancer driver genes particularly when the fraction of selected nucleotides on a gene is small. Using the combined test, we discovered a catalogue of well-known cancer driver genes as well as a long tail of novel candidate cancer genes with mutation frequencies as low as 1% and functional supporting evidence. The signal of negative selection is very subtle, but is detectable in several cancer types and in a pan-cancer data set. It is enriched in cell-essential genes identified in a CRISPR knockout screen, as well as in genes with reported roles in cancer.

Open to predoctoral UPF students

If you are interested in participating please send an email to cristina.mendez@irbbarcelona.org