Cancer driver gene identification method based on machine learning and various statistic principles

Abstract

The invention discloses a cancer driver gene identification method based on machine learning and various statistic principles. The method includes following steps: (1), organizing data into a standardformat; (2), calculating a background variation rate; (3), statistically testing cancer driver gene; (4), performing Monte-Carlo simulation on statistic distribution; (5), adjusting P value. The background variation rate of each sample, gene and mutation types and influence of various mutation types on protein functions are taken into consideration, and score testing is adopted to judge the driver gene, so that the method has high robustness and is widely suitable for cancer of various types; sensitivity and specificity are balanced well, more driver genes can be detected while low false positive rate can be maintained. The method is of important significance in looking for potential loci for cancer treatment and developing anticancer drug.

Description

technical field [0001] The invention belongs to the interdisciplinary field of bioinformatics and cancer medicine, and relates to a cancer driver gene identification method using machine learning and various statistical methods. Background technique [0002] Most cancers are diseases caused by mutations in somatic cells. Driver genes are factors that directly lead to the occurrence and development of cancer. On the other hand, there is no direct relationship between passenger genes and cancer, so it is necessary to identify driver genes. Several major tumor sequencing projects in the world, such as Cancer Genome Atlas Project (TCGA), International Cancer Genome Consortium Project (ICGC) and Clinical Application Research to General Effective Therapy Project (TARGET), have established comprehensive catalogs of somatic mutations in various types of cancer. A major goal of these sequencing projects is to identify driver genes that cause cancer. Finding cancer driver genes can ...

AI Technical Summary

Problems solved by technology

[0004] However, the above-mentioned bioinformatics tools still have some shortcomings. First, these algorithms have not achieved a good balance between sensitivity and specificity, that is, some algorithms have high sensitivity but low specificity, or high specificity and low sensitivity; secondly, These methods lack robustness to different tumor types, that is, for some tumor types, the method performs well and finds many reliable driver genes, but shows poor performance for other tumor types

Images