Shape-based retrieval of CNV regions in read coverage data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Copy number variations (CNVs) are significant sources of human genetic diversity and disease.
• Accurate detection of CNVs is crucial for genomic studies and clinical applications.
• Existing CNV detection methods struggle with noise in real-world sequencing data.

Purpose of the Study:

• To develop a novel and robust method for detecting copy number variations (CNVs).
• To improve the precision and sensitivity of CNV detection, especially in low-coverage and noisy genomic data.
• To enhance the performance of CNV detection compared to conventional approaches.

Main Methods:

• The CNV_shape method utilizes read coverage shape variations for CNV detection.
• It employs mean shift transform for preliminary CNV estimation and mean slope transform for refining candidate regions.
• A merging score based on transform properties is used to identify final CNVs.

Main Results:

• CNV_shape successfully detected small CNVs (> 1 kbp) from low-coverage data (> 1.7x).
• The method demonstrated significant performance improvements, ranging from 8.18% to 87.90%, over conventional methods.
• CNV_shape proved effective in mitigating experimental and biological noise in real human genome data.

Conclusions:

• CNV_shape offers a highly effective approach for CNV detection across various sizes and types.
• The method's robustness to noise makes it suitable for analyzing real-world genomic datasets.
• This advancement has implications for improving genomic analysis and understanding genetic variation.