Comparison of sequencing based CNV discovery methods using monozygotic twin quartets.

Area of Science:

• Genomics
• Bioinformatics

Background:

• High-throughput sequencing presents challenges in identifying copy-number variations (CNVs).
• CNVs are structural variations involving large genomic fragments (>1 kilobase).
• Evaluating CNV calling methods is crucial for accurate genomic analysis.

Purpose of the Study:

• To compare the consistency and accuracy of different CNV calling algorithms.
• To assess algorithm performance using whole-genome sequencing data from identical twin pairs.
• To determine the effectiveness of combining CNV call sets.

Main Methods:

• Utilized whole-genome sequencing data from 9 quartets of identical twins.
• Applied multiple CNV calling algorithms to the sequencing data.
• Analyzed CNV calls for consistency, Mendelian inheritance, and breakpoint resolution.

Main Results:

• ERDS and CNVnator demonstrated the highest inherited CNV rates (0.74 and 0.70).
• Filtering familial inconsistencies revealed high false positive rates for CNVer and Breakdancer.
• Combining ERDS and CNVnator calls yielded the highest inherited CNV rate (81%) and superior breakpoint resolution.

Conclusions:

• ERDS and CNVnator are recommended for CNV calling from whole-genome sequencing data due to their consistency and accuracy.
• Combining calls from ERDS and CNVnator enhances CNV genotyping pipelines.
• The study highlights the value of using identical twins to assess CNV caller performance.