Improving structural variant clustering to reduce the negative effect of the breakpoint uncertainty problem

Area of Science:

• Genomics
• Bioinformatics
• Population Genetics

Background:

• Structural variants (SVs) are a key source of genetic variation.
• Breakpoint uncertainty in SV detection using short-read sequencing hinders population analyses.
• Current methods often cluster SVs to mitigate uncertainty before merging sample data.

Purpose of the Study:

• To compare dissimilarity measures for SV clustering.
• To evaluate SV clustering performance based on Mendelian inheritance errors (MIE), kinship prediction, and Hardy-Weinberg equilibrium.
• To introduce a new measure of dataset consistency and a constrained clustering method.

Main Methods:

• Comparison of two dissimilarity measures for SV clustering.
• Analysis of Mendelian-inconsistent SV clusters as a measure of dataset consistency.
• Development and application of a constrained clustering method.

Main Results:

• The dissimilarity measure based on breakpoint distance slightly outperformed overlap-based measures.
• Constrained clustering demonstrated superior performance across all evaluated metrics.
• The new method effectively identified Mendelian-inconsistent SV clusters.

Conclusions:

• Constrained clustering offers the most robust approach for structural variant analysis.
• The choice of dissimilarity measure is less critical when using constrained clustering.
• Improved SV clustering enhances the reliability of genetic variation studies.