PreGoLoF: Predict the Gain or Loss of Functions

Release 1.0

 

BACKGROUND
Functional gain or loss of somatic mutations in cancer
Considering cancer initiation and progression as an evolutionary process, the cancer associated gene mutations, or “driver mutations”, are selected to promote the cancer tissue’s survival by increasing the capability of the tissue to overcome certain stressors in its microenvironments. Gain or loss of function of multiple cancer associated mutations including TP53, KRAS, VHL, and PI3KCA are well acknowledged. Such observations indicate that same mutation may serve different roles in different cancer tissues. On the other hand, same mutation may be selected by different reasons. It is worth to note that most of the “driver” mutations, especially the suppressor genes as TP53 and VHL always involve in large set of pathways with distinct cellular functions, suggesting possible heterogeneous gain or loss of functions can be led by the mutation. Unfortunately, such heterogeneous gain or loss of functions are excluded in the assumption of most of the current mutation identification method. To our best knowledge, this work is the first computational framework focusing on comprehensive prediction of possible functional changes of somatic mutations in cancer.
Figure 1 illustrates the multiple function of APC gene. Mutations on specific region may lead to certain functional gain or loss.

Figure 1. Multiple functions of APC gene (mutation). The figure is from: Koji Aoki, Makoto M. Taketo. Adenomatous polyposis coli (APC): a multi-functional tumor suppressor gene. Journal of Cell Science 2007 120: 3327-3335; doi: 10.1242/jcs.03485

Possible roles of “passenger” mutations in cancer.
We have conducted an analysis of two sets of genomes of precancerous and cancerous colon tissues in the public domain. The first set consists of 20 precancerous and early cancer samples: one polyp having 4 mutations, eight small adenoma samples harboring 272 mutations, eight large adenoma samples having 344 mutations and three adenocarcinoma samples with 198 mutations. The second consists of 131 samples: 18 stage-1 adenocarcinoma tissues with 1,439 mutations, 47 stage-2 samples with 3,683 mutations, 43 stage-3 samples harboring 3,657 mutations and 23 stage-4 samples having 2,061 mutations. According to the authors of the original studies, the first set has nine driver mutations and the second has 32. We have analyzed only the passenger mutations predicted by the original authors to find out if useful information can be derived from the “passenger” mutations.
We first identified pathways enriched by these mutations among tissues in each disease stage, and made the following observations all with high statistical significance: (i) mutations in small adenoma enrich functional groups related to the composition of the ECM, cell-ECM interaction, cell-cell adhesion, cell morphology and cell cycle control; (ii) mutations in large adenoma enrich the following in addition to (i): EGF-like domain, ABC transporters, cadherin, and actin binding; (iii) mutations in stage-1 adenocarcinoma enrich functional groups associated with ion transporter, plasma membrane, immunoglobulin and complement control; in addition, they also enrich functional groups associated with cell adhesion, ECM composition, cytoskeletal structure and ATP binding, which are true for mutations across stage 2-4 samples; and (iv) mutations in stages 2-4 samples enrich those related tocytoskeletal reorganization; cell motion; differentiation; embryonic development; and a tyrosine kinase.