Improving Copy Number Variant Detection from Sequencing Data with a Combination of Programs and a Predictive Model.

Area of Science:

• Genomics
• Bioinformatics
• Genetic Disorders

Background:

• Copy number variants (CNVs) analysis from gene panel massively parallel sequencing (MPS) data is challenging due to less developed bioinformatics tools.
• Accurate CNV detection is crucial for diagnosing genetic disorders, including neuromuscular disorders.

Purpose of the Study:

• To develop and validate an efficient bioinformatics pipeline for CNV detection from gene panel MPS data.
• To evaluate the performance of different CNV detection programs and their combinations.
• To establish a statistical model for improving the accuracy and standardization of CNV detection.

Main Methods:

• In silico generation of CNVs in MPS gene panel data.
• Analysis of in silico CNVs using four complementary programs: CoNIFER, XHMM, ExomeDepth, and CODEX.
• Training and validation of a logistic regression model using CNV detection scores from the four programs.

Main Results:

• A combination of all four CNV detection programs yielded more sensitive results than individual programs or other combinations.
• The logistic regression model incorporating scores from all four programs demonstrated superior performance in validation.
• No single program achieved sufficient accuracy for detecting all CNV sizes and types.

Conclusions:

• A combination of carefully selected bioinformatics tools is essential for maximizing CNV detection accuracy.
• A statistical model aids in streamlining and standardizing the filtering and annotation of detected CNVs.
• The developed pipeline offers an efficient approach for CNV analysis in gene panel MPS data for neuromuscular disorders.