MLFSdel: An accurate approach to discover genome deletions

摘要

Genome deletions are one of the common types of structural variations. The discovery of deletions has become an important research field in SVs detection of genome sequences. At present, the existing methods have their own limitations, and these methods are also insufficient in precision and sensitivity. Hence, improving the detecting efficiency has become a critical target in subsequent research. In this paper, we developed a method, namely MLFSdel. Essentially, MLFSdel employs four machine learning models and implements a novel feature selection strategy. By eliminating the features having the negative effect on the overall classification results, the proposed method improves the precision and sensitivity in comparison to four previous methods for detecting deletions. In addition, it further proves that the feature-based machine learning methods are applicable to detect genome deletions.