Complex indel is a special kindof genomic structural variations. Complex indel usually presents as one or more gene fragments inserted into places where deletion mutation occurs, resulting in complex insertion and deletion mutation. The detection of complex indel and genotyping are of great scientific research value in the field of disease association analysis. In recent years, it has been one of the hot spots in the field of bioinformatics.This project focuses on the genotyping algorithm for complex indel.
In this project, we implemented a machine learning algorithm based on relevance vector machine to realize the genotyping for complex indel. We have completed sequence alignment, data preprocessing, feature analysis and extraction, training of relevance vector machine in our work. In the genotyping algorithm, we extracted the abnormal read distance, the number of partially matched reads, the number of singly matched reads, the number of perfectly matched reads, and the reads frequency as features. Based on the extracted features and the known genotypes, we trained the relevance vector machine classifier, and transformed the genotyping problem into the classification problem, which can be solved bymachine learning algorithms.