A systematic evaluation of copy number alterations detection methods on real SNP array and deep sequencing data

Area of Science:

• Genomics
• Cancer Research
• Bioinformatics

Background:

• Copy Number Alterations (CNAs) are crucial in cancer development.
• Accurate CNA detection is vital for cancer genomics.
• Existing evaluation studies lack consensus due to limited data, confusing researchers on method selection.

Purpose of the Study:

• To comprehensively evaluate the performance of 12 widely used Copy Number Alteration (CNA) detection methods.
• To provide a clear comparison of these methods across different data types and scenarios.
• To offer guidance for selecting the most appropriate CNA detection tool.

Main Methods:

• Evaluation of 12 typical CNA detection methods using a large-scale dataset from the CAGEKID consortium.
• Dataset includes SNP array data (94 samples) and whole genome sequencing data (10 samples).
• Methods assessed on SNP array data, sequencing data with matched tumor-normal samples, and sequencing data from single tumor samples.

Main Results:

• Detailed performance comparison of 12 CNA detection methods, highlighting their strengths and weaknesses.
• Ranking of SNP-based methods followed by comparison with matched and single-sample sequencing-based methods.
• Analysis of preprocessing, recall rate, Jaccard index, and segmentation characteristics.

Conclusions:

• The study elucidates the comparative performance of 12 CNA detection methods.
• Methodological explanations are provided for observed performance differences.
• A guiding principle for selecting CNA detection methods based on specific requirements is presented.