Improving CNV Detection Performance in Microarray Data Using a Machine Learning-Based Approach

Area of Science:

• Genomics
• Medical Genetics
• Bioinformatics

Background:

• Copy number variation (CNV) is a major cause of structural genomic variation and associated disorders.
• Accurate analysis of neonatal CNVs is critical for managing chromosomal disabilities.
• Genomic waves can compromise the precision of CNV detection.

Purpose of the Study:

• To develop and validate a novel machine learning-based method for accurate neonatal CNV analysis.
• To address the challenge of genomic waves affecting CNV detection accuracy.
• To identify CNVs and associated genetic disorders in a large cohort of Korean newborns.

Main Methods:

• Development of a new method utilizing a modified log R ratio to counteract genomic wave influences.
• Application of a machine learning approach for enhanced CNV detection.
• Validation using samples with known CNVs and comparison with Next-Generation Sequencing (NGS) data.

Main Results:

• The new method demonstrated superior performance in identifying CNVs compared to traditional log R ratio methods.
• Analysis of 16,046 Korean newborns revealed CNVs associated with 39 genetic disorders in 342 cases.
• Joubert syndrome 4 was the most frequently identified CNV-related disorder.

Conclusions:

• The developed method, employing a genome-wide single nucleotide polymorphism array with wave offset, is effective for neonatal CNV identification.
• Accurate CNV screening facilitates the identification of disease susceptibilities and chromosomal disease etiologies.
• This approach can significantly advance the diagnosis and management of CNV-related genetic disorders in newborns.