InvBFM

InvBFM enhances detection of genomic inversions from high-throughput sequencing data by applying feature mining and a support vector machine classifier to distinguish true inversions.

Key Features:

• Inversion detection methodology: Employs feature mining combined with machine learning to improve precision and sensitivity of inversion detection.
• Integration of existing methods: Aggregates candidate inversions from established methods that use paired-end reads, split-mapped reads, and assembly approaches.
• Feature extraction: Extracts relevant features from candidate inversions to enable accurate classification.
• Machine learning model: Utilizes a trained support vector machine (SVM) classifier to discern true inversions from false positives.
• Performance: Demonstrated higher precision and sensitivity on real sequence data from the 1000 Genomes Project compared to existing tools.
• Data preprocessing: Includes Shell programs fixRealData and fixSimuData for input data preprocessing.
• Implementation: Implemented primarily in Python with additional Shell scripting components.

Scientific Applications:

• Structural variation studies: Detection of genomic inversions for research on structural variation in genomes.
• Genetics and evolutionary biology: Supports studies on genetic variation and evolutionary analyses involving inversions.
• Disease research: Facilitates identification of inversion events that may associate with phenotypic traits or disorders.

Methodology:

Aggregates inversion candidates from paired-end, split-mapped, and assembly-based methods; performs feature extraction and feature mining; classifies candidates using a trained support vector machine (SVM); includes Shell scripts fixRealData and fixSimuData for preprocessing; implemented in Python with Shell components.