Mutation-profile-based methods for understanding selection forces in cancer somatic mutations: a comparative analysis

U2AF1 mutations rescue deleterious exon skipping induced by KRAS mutations

The mechanisms by which somatic mutations of splicing factors, such as U2AF1 S34F in lung adenocarcinoma, contribute to cancer pathogenesis are not well understood. Here, we used prime editing to modify the endogenous U2AF1 gene in lung adenocarcinoma cells and assessed the resulting impact on alternative splicing. These analyses identified KRAS as a key target modulated by U2AF1 S34F . One specific KRAS mutation, G12S, generates a cryptic U2AF1 binding site that leads to skipping of KRAS exon 2 and generation of a non-functional KRAS transcript. Expression of the U2AF1 S34F mutant reverts this exon skipping and restores KRAS function. Analysis of cancer genomes reveals that U2AF1 S34F mutations are enriched in KRAS G12S -mutant lung adenocarcinomas. A comprehensive analysis of splicing factor/oncogene mutation co-occurrence in cancer genomes also revealed significant co-enrichment of KRAS Q61R and U2AF1 I24T mutations. Experimentally, KRAS Q61R mutation leads to KRAS exon 3 skipping, which in turn can be rescued by the expression of U2AF1 I24T . Analysis of genomic and clinical patient data suggests that both U2AF1 mutations occur secondary to KRAS mutation and are associated with decreased overall patient survival. Our findings provide evidence that splicing factor mutations can rescue splicing defects caused by oncogenic mutations. More broadly, they demonstrate a dynamic process of cascading selection where mutational events are positively selected in cancer genomes as a consequence of earlier mutations.

Neutral evolution of rare cancer mutations in the computer and the clinic

A distinct model of neutral evolution of rare cancer mutations is described and contrasted with models relying on the infinite sites approximation (that a specific mutation arises in only one cell at any instant). An explosion of genetic diversity is predicted at clinical cell numbers and may explain the progressive refractoriness of cancers during a clinical course. The widely used infinite sites assumption may not be applicable for clinical cancers.

Evolutionary dependency of cancer mutations in gene pairs inferred by nonsynonymous-synonymous mutation ratios

BACKGROUND

Determining the impact of somatic mutations requires understanding the functional relationship of genes acquiring mutations; however, it is largely unknown how mutations in functionally related genes influence each other.

METHODS

We employed non-synonymous-to-synonymous or dNdS ratios to evaluate the evolutionary dependency (ED) of gene pairs, assuming a mutation in one gene of a gene pair can affect the evolutionary fitness of mutations in its partner genes as mutation context. We employed PanCancer- and tumor type-specific mutational profiles to infer the ED of gene pairs and evaluated their biological relevance with respect to gene dependency and drug sensitivity.

RESULTS

We propose that dNdS ratios of gene pairs and their derived cdNS (context-dependent dNdS) scores as measure of ED distinguishing gene pairs either as synergistic (SYN) or antagonistic (ANT). Mutation contexts can induce substantial changes in the evolutionary fitness of mutations in the paired genes, e.g., IDH1 and IDH2 mutation contexts lead to substantial increase and decrease of dNdS ratios of ATRX indels and IDH1 missense mutations corresponding to SYN and ANT relationship with positive and negative cdNS scores, respectively. The impact of gene silencing or knock-outs on cell viability (genetic dependencies) often depends on ED, suggesting that ED can guide the selection of candidates for synthetic lethality such as TCF7L2-KRAS mutations. Using cell line-based drug sensitivity data, the effects of targeted agents on cell lines are often associated with mutations of genes exhibiting ED with the target genes, informing drug sensitizing or resistant mutations for targeted inhibitors, e.g., PRSS1 and CTCF mutations as resistant mutations to EGFR and BRAF inhibitors for lung adenocarcinomas and melanomas, respectively.

CONCLUSIONS

We propose that the ED of gene pairs evaluated by dNdS ratios can advance our understanding of the functional relationship of genes with potential biological and clinical implications.

Unraveling the Drivers of Tumorigenesis in the Context of Evolution: Theoretical Models and Bioinformatics Tools

Cancer originates from somatic cells that have accumulated mutations. These mutations alter the phenotype of the cells, allowing them to escape homeostatic regulation that maintains normal cell numbers. The emergence of malignancies is an evolutionary process in which the random accumulation of somatic mutations and sequential selection of dominant clones cause cancer cells to proliferate. The development of technologies such as high-throughput sequencing has provided a powerful means to measure subclonal evolutionary dynamics across space and time. Here, we review the patterns that may be observed in cancer evolution and the methods available for quantifying the evolutionary dynamics of cancer. An improved understanding of the evolutionary trajectories of cancer will enable us to explore the molecular mechanism of tumorigenesis and to design tailored treatment strategies.

MODIG: Integrating Multi-Omics and Multi-Dimensional Gene Network for Cancer Driver Gene Identification based on Graph Attention Network Model

MOTIVATION

Identifying genes that play a causal role in cancer evolution remains one of the biggest challenges in cancer biology. With the accumulation of high-throughput multi-omics data over decades, it becomes a great challenge to effectively integrate these data into the identification of cancer driver genes.

RESULTS

Here, we propose MODIG, a graph attention network (GAT)-based framework to identify cancer driver genes by combining multi-omics pan-cancer data (mutations, copy number variants, gene expression, and methylation levels) with multi-dimensional gene networks. First, we established diverse types of gene relationship maps based on protein-protein interactions (PPI), gene sequence similarity, KEGG pathway co-occurrence, gene co-expression patterns, and Gene Ontology (GO). Then, we constructed a multi-dimensional gene network consisting of approximately 20,000 genes as nodes and five types of gene associations as multiplex edges. We applied a GAT to model within-dimension interactions to generate a gene representation for each dimension based on this graph. Moreover, we introduced a joint learning module to fuse multiple dimension-specific representations to generate general gene representations. Finally, we used the obtained gene representation to perform a semi-supervised driver gene identification task. The experiment results show that MODIG outperforms the baseline models in terms of area under precision-recall curves (AUPR) and area under the receiver operating characteristic curves (AUROC).

AVAILABILITY

The MODIG program is available at https://github.com/zjupgx/modig. The code and data underlying this article are also available on Zenodo, at https://doi.org/10.5281/zenodo.7057241.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

dN/dS-H, a New Test to Distinguish Different Selection Modes in Protein Evolution and Cancer Evolution

One of the most popular measures in the analysis of protein sequence evolution is the ratio of nonsynonymous distance (d_N) to synonymous distance (d_S). Under the assumption that synonymous substitutions in the coding region are selectively neutral, the d_N/d_S ratio can be used to statistically detect the adaptive evolution (or purifying selection) if d_N/d_S > 1 (or d_N/d_S < 1) significantly. However, due to strong structural constraints and/or variable functional constraints imposed on amino acid sites, most encoding genes in most species have demonstrated d_N/d_S < 1. Consequently, the statistical power for testing d_N/d_S = 1 may be insufficient to distinguish between different selection modes. In this paper, we propose a more powerful test, called d_N/d_S-H, in which a new parameter H, a relative measure of rate variation among sites, was introduced. Given the condition of strong purifying selections at some sites, the d_N/d_S-H model predicts d_N/d_S = 1-H for neutral evolution, d_N/d_S < 1-H for nearly neutral selection, and d_N/d_S > 1-H for adaptive evolution. The potential of this new method for resolving the neutral-adaptive debates is illustrated by the protein sequence evolution in vertebrates, Drosophila and yeasts, as well as somatic cancer evolution (specialized as the C_N/C_S-H test).

Use of Publication Dynamics to Distinguish Cancer Genes and Bystander Genes

de Magalhães has shown recently that most human genes have several papers in PubMed mentioning cancer, leading the author to suggest that every gene is associated with cancer, a conclusion that contradicts the widely held view that cancer is driven by a limited number of cancer genes, whereas the majority of genes are just bystanders in carcinogenesis. We have analyzed PubMed to decide whether publication metrics supports the distinction of bystander genes and cancer genes. The dynamics of publications on known cancer genes followed a similar pattern: seminal discoveries triggered a burst of cancer-related publications that validated and expanded the discovery, resulting in a rise both in the number and proportion of cancer-related publications on that gene. The dynamics of publications on bystander genes was markedly different. Although there is a slow but continuous time-dependent rise in the proportion of papers mentioning cancer, this phenomenon just reflects the increasing publication bias that favors cancer research. Despite this bias, the proportion of cancer papers on bystander genes remains low. Here, we show that the distinctive publication dynamics of cancer genes and bystander genes may be used for the identification of cancer genes.

CanDriS: posterior profiling of cancer-driving sites based on two-component evolutionary model

Current cancer genomics databases have accumulated millions of somatic mutations that remain to be further explored. Due to the over-excess mutations unrelated to cancer, the great challenge is to identify somatic mutations that are cancer-driven. Under the notion that carcinogenesis is a form of somatic-cell evolution, we developed a two-component mixture model: while the ground component corresponds to passenger mutations, the rapidly evolving component corresponds to driver mutations. Then, we implemented an empirical Bayesian procedure to calculate the posterior probability of a site being cancer-driven. Based on these, we developed a software CanDriS (Cancer Driver Sites) to profile the potential cancer-driving sites for thousands of tumor samples from the Cancer Genome Atlas and International Cancer Genome Consortium across tumor types and pan-cancer level. As a result, we identified that approximately 1% of the sites have posterior probabilities larger than 0.90 and listed potential cancer-wide and cancer-specific driver mutations. By comprehensively profiling all potential cancer-driving sites, CanDriS greatly enhances our ability to refine our knowledge of the genetic basis of cancer and might guide clinical medication in the upcoming era of precision medicine. The results were displayed in a database CandrisDB (http://biopharm.zju.edu.cn/candrisdb/).

Use of signals of positive and negative selection to distinguish cancer genes and passenger genes

A major goal of cancer genomics is to identify all genes that play critical roles in carcinogenesis. Most approaches focused on genes positively selected for mutations that drive carcinogenesis and neglected the role of negative selection. Some studies have actually concluded that negative selection has no role in cancer evolution. We have re-examined the role of negative selection in tumor evolution through the analysis of the patterns of somatic mutations affecting the coding sequences of human genes. Our analyses have confirmed that tumor suppressor genes are positively selected for inactivating mutations, oncogenes, however, were found to display signals of both negative selection for inactivating mutations and positive selection for activating mutations. Significantly, we have identified numerous human genes that show signs of strong negative selection during tumor evolution, suggesting that their functional integrity is essential for the growth and survival of tumor cells.

Negative selection in tumor genome evolution acts on essential cellular functions and the immunopeptidome

Background

Natural selection shapes cancer genomes. Previous studies used signatures of positive selection to identify genes driving malignant transformation. However, the contribution of negative selection against somatic mutations that affect essential tumor functions or specific domains remains a controversial topic.

Results

Here, we analyze 7546 individual exomes from 26 tumor types from TCGA data to explore the portion of the cancer exome under negative selection. Although we find most of the genes neutrally evolving in a pan-cancer framework, we identify essential cancer genes and immune-exposed protein regions under significant negative selection. Moreover, our simulations suggest that the amount of negative selection is underestimated. We therefore choose an empirical approach to identify genes, functions, and protein regions under negative selection. We find that expression and mutation status of negatively selected genes is indicative of patient survival. Processes that are most strongly conserved are those that play fundamental cellular roles such as protein synthesis, glucose metabolism, and molecular transport. Intriguingly, we observe strong signals of selection in the immunopeptidome and proteins controlling peptide exposition, highlighting the importance of immune surveillance evasion. Additionally, tumor type-specific immune activity correlates with the strength of negative selection on human epitopes.

Conclusions

In summary, our results show that negative selection is a hallmark of cell essentiality and immune response in cancer. The functional domains identified could be exploited therapeutically, ultimately allowing for the development of novel cancer treatments.

Electronic supplementary material

The online version of this article (10.1186/s13059-018-1434-0) contains supplementary material, which is available to authorized users.