SVScore: an impact prediction tool for structural variation.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Structural variations (SVs) are significant contributors to human genetic diversity and disease.
• Accurate prediction of SV impact is crucial for understanding genetic disorders.
• Existing methods for SV impact prediction have limitations.

Purpose of the Study:

• To introduce SVScore, a novel tool for in silico prediction of structural variation (SV) impact.
• To develop a method that aggregates per-base single nucleotide polymorphism (SNP) pathogenicity scores for SVs.
• To evaluate SVScore's performance against alternative prediction methods.

Main Methods:

• SVScore aggregates per-base SNP pathogenicity scores across genomic intervals for each SV.
• The method incorporates variant type, gene features, and positional uncertainty into its scoring.
• Implementation in Perl, freely available under the MIT license.

Main Results:

• High-scoring SVs exhibit an allele frequency spectrum skewed toward lower frequencies, indicating purifying selection.
• SVScore demonstrates superior performance in identifying deleterious variants compared to alternative methods.
• Duplications appear to be under stronger selection than deletions, with a comparable number of pathogenic SVs and SNPs in the human population.

Conclusions:

• SVScore provides an effective tool for predicting the impact of structural variations.
• The findings highlight the significant role of SVs in human genetics and disease.
• The study suggests that both SVs and SNPs contribute substantially to pathogenic genetic variation.