Somatic Mutation Detection Using Ensemble of Machine Learning

The continuous advancement of next-generation sequencing (NGS) technology enables researchers to detect somatic mutations, significantly enhancing the accuracy of identifying somatic mutations from NGS data. With the continuous advancement of Machine Learning (ML) technology, researchers have gained more confidence in utilizing this technology for data prediction. This article proposes the combination of the Tree-structured Parzen Estimator (TPE) algorithm with the Support Vector Machines (SVM) algorithm for detecting somatic mutations in matched tumor and normal paired sequencing data. The method is applied to real biological data from exome capture data and whole-genome shotgun data. The results indicate a significant improvement in the detectability of somatic mutations using the proposed integrated approach compared to the conventional methods.