EigenDel

EigenDel detects genomic deletions from high-throughput sequence data to identify structural variation (SV) ranging from 50 base pairs to approximately 3 megabases that impacts genetic diversity and disease.

Key Features:

• Signal Utilization: Leverages discordant read-pairs, read depth, split reads and clipped reads, with primary focus on discordant read-pairs and clipped reads for initial deletion candidate identification.
• Unsupervised Clustering: Applies unsupervised learning to cluster similar deletion candidates to distinguish true deletions from noise.
• Deletion Calling Strategy: Calls true deletions within each cluster using a specialized approach to minimize false positives.
• SV Size Range: Targets structural variant deletions from 50 base pairs to approximately 3 megabases.
• Performance Characteristics: Balances accuracy and sensitivity and reduces bias to enable detection in low coverage sequence data.

Scientific Applications:

• Genetic research: Identification of deletions relevant to genetic diversity and variant discovery.
• Disease association studies: Detection of deletions that may contribute to disease phenotypes.
• Evolutionary biology: Analysis of deletion polymorphisms affecting evolution and population variation.
• Low-coverage sequencing studies: Reliable deletion detection in datasets with low sequencing coverage.

Methodology:

Initial candidate identification using discordant read-pairs and clipped reads; clustering of candidates with unsupervised learning; and calling true deletions within each cluster.