Abstract 1638: Whole-exome sequencing and protein interaction networks to prioritize candidate genes for cutaneous melanoma susceptibility

Abstract

Abstract Background: Known cutaneous malignant melanoma (CMM) genes account for melanoma risk in less than 40% of melanoma-prone families, suggesting the existence of additional risk genes or other modifiers. Whole exome sequencing (WES) of high-risk families usually results in the identification of a large number of potential disease-causing genes. However, prioritizing these genes and finding the true disease-causing ones have been challenging. We hypothesize that physically interacting proteins might have a similar effect on cell function and therefore, might produce a shared phenotype or increase the risk when mutated. To test this hypothesis, we applied a network approach integrating WES and protein-protein interaction (PPI) data to detect and prioritize rare variants in melanoma-prone families. Methods: We conducted WES analysis on germline DNA of 212 patients from 45 CMM families (2-5 affected people in a family) without known mutations. Variants that were rare (<1% in public and our in-house control datasets), loss-of-function or missense predicted to be deleterious, and showed complete disease co-segregation (present in all affected members within a family) were included in the network analysis. For this analysis, we only focused on genes that interact with known CMM genes (gene list in Goldstein AM, et al. Hum Mol Genet, 2017) which we used as seed proteins. We then applied a degree-aware algorithm (DADA) to rank the set of candidate genes from WES with respect to the set of known CMM genes using human PPI networks. The goal of DADA is to prioritize a candidate set of genes based on their association level with genes known to be related to the disease. Network proximity calculations of the candidates with respect to the seed set were based on Random Walk with Restarts method. We used high quality and scored human proteins interactions collected from InWeb_IM. In the primary analysis, the network contains 625,641 interactions, aggregated from 8 source databases and spanning 87% of reviewed human proteins. We also conducted additional analyses with CMM PPI subnetworks generated using high confidence interaction data from different sources to compare results. Results: The WES analysis described above identified 546 genes that harbored rare variants, likely affect protein function and co-segregated in affected members in CMM families. After applying DADA across all networks, the rankings of these genes were fairly consistent. Top genes included both known CMM genes (such as MC1R, CDKN2B, KITLG, and TINF2) and genes that have not yet been associated with CMM susceptibility (such as FOXK1, AR, LEF1, RGL4, PABPC1, CHEK2, CTBP2, and BLM). These results highlight the importance of known CMM genes and their networks in CMM susceptibility, and demonstrate the potential value of network approaches in gene prioritization. Further evaluation of novel genes is in progress. Citation Format: Sally Yepes, Margaret Tucker, Hela Koka, Kristine Jones, Aurelie Vogt, Laurie Burdette, Wen Luo, Bin Zhu, Meredith Yeager, Belynda Hicks, Neal D. Freedman, Stephen J. Chanock, Alisa M. Goldstein, Xiaohong R. Yang. Whole-exome sequencing and protein interaction networks to prioritize candidate genes for cutaneous melanoma susceptibility [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 1638.