LUMPY: a probabilistic framework for structural variant discovery.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Comprehensive structural variation (SV) discovery from whole genome sequencing (WGS) data relies on integrating multiple detection signals.
• Existing SV detection algorithms often use signals in isolation or sequentially, limiting comprehensive discovery.
• Technical challenges hinder the joint integration of diverse SV detection signals.

Purpose of the Study:

• To present LUMPY, a novel framework for the joint integration of multiple SV detection signals across multiple samples.
• To improve the sensitivity of structural variation discovery, particularly in challenging datasets.
• To provide a validated set of structural variation breakpoints from a human genome.

Main Methods:

• Development of LUMPY, a novel computational framework for structural variation detection.
• Joint integration of multiple SV detection signals: read-pair, split-read, read-depth, and prior knowledge.
• Application of LUMPY to whole genome sequencing data, including low-coverage samples.

Main Results:

• LUMPY demonstrates improved sensitivity for structural variation discovery compared to existing methods.
• Enhanced performance is observed particularly with low-coverage sequencing data or low intra-sample variant allele frequency.
• A set of 4,564 validated structural variation breakpoints from the NA12878 human genome was identified.

Conclusions:

• LUMPY offers a robust and sensitive approach for comprehensive structural variation discovery.
• The framework effectively integrates multiple signals, overcoming limitations of previous algorithms.
• LUMPY advances the field of genomic variation analysis, providing valuable insights from WGS data.