KNNCNV: A K-Nearest Neighbor Based Method for Detection of Copy Number Variations Using NGS Data

Area of Science:

• Genomics
• Bioinformatics
• Cancer Research

Background:

• Copy number variation (CNV) is a genomic mutation associated with human cancers.
• Accurate CNV detection is vital for cancer mutation analysis, diagnosis, and treatment.
• Next-generation sequencing (NGS) enables base-level CNV detection, but challenges remain due to data complexity.

Purpose of the Study:

• To introduce KNNCNV, a novel K-Nearest Neighbor based method for detecting CNVs from NGS data.
• To address the challenges in accurate CNV detection, particularly for local variations.
• To provide an automated threshold-free approach for CNV identification.

Main Methods:

• KNNCNV assigns an outlier score to genome segments based on k-nearest neighbor distances.
• The variational Bayesian Gaussian mixture model (VBGMM) converts scores to binary labels without user-defined thresholds.
• Performance was evaluated using simulated and real sequencing data, compared against existing methods.

Main Results:

• KNNCNV demonstrates improved performance in CNV detection compared to peer methods.
• The method excels in identifying local CNVs that might be obscured in surrounding regions.
• Achieved superior F1-scores in experimental evaluations.

Conclusions:

• KNNCNV offers a powerful and accurate alternative for CNV detection using NGS data.
• The method's approach to outlier scoring and automated thresholding enhances detection capabilities.
• KNNCNV has the potential to advance cancer genomics research and clinical applications.