InvBFM: finding genomic inversions from high-throughput sequence data based on feature mining.

Area of Science:

• Genomics
• Bioinformatics

Background:

• Genomic inversions are significant structural variations (SVs) with crucial biological functions.
• Accurate detection of inversions from high-throughput sequencing data remains challenging due to complex genomic regions and limitations of existing tools.
• Current inversion detection methods based on paired-end reads, split-mapped reads, or assembly exhibit suboptimal precision and sensitivity.

Purpose of the Study:

• To develop a novel and more accurate method for calling genomic inversions.
• To address the limitations in sensitivity and precision of existing inversion detection tools.

Main Methods:

• InvBFM utilizes a feature mining approach to identify genomic inversions.
• The method integrates candidate inversions from existing tools.
• A support vector machine (SVM) classifier is trained on extracted features to accurately call true inversions.

Main Results:

• InvBFM demonstrates superior performance compared to existing inversion detection tools.
• The method achieves higher accuracy by combining feature mining with a machine learning model.
• Validation was performed using real sequence data from the 1000 Genomes Project.

Conclusions:

• The combination of feature mining and machine learning significantly improves inversion calling accuracy.
• InvBFM offers a more effective solution for detecting genomic inversions.
• The InvBFM software is publicly available for use in genomic sequence analysis.