Benchmarking copy number aberrations inference tools using single-cell multi-omics datasets

Area of Science:

• Genomics
• Computational Biology
• Cancer Research

Background:

• Copy number alterations (CNAs) are key genomic variations driving cancer initiation and progression.
• Single-cell RNA sequencing (scRNA-seq) enables CNA inference, but method performance lacks comprehensive benchmarking.

Purpose of the Study:

• To comprehensively evaluate and compare the performance of five state-of-the-art computational methods for inferring CNAs from scRNA-seq data.
• To provide guidelines for selecting appropriate CNA inference tools based on specific research needs and datasets.

Main Methods:

• Evaluated five leading computational methods for CNA inference from scRNA-seq data.
• Assessed performance based on tumor vs. normal cell classification, CNA profile accuracy, tumor subclone inference, and aneuploidy identification.
• Investigated the impact of referencing settings, tumor microenvironment cell inclusion, tumor type, and purity on method performance.

Main Results:

• Numbat generally outperformed other methods across most evaluation criteria.
• CopyKAT showed excellent performance when using only the expression matrix.
• SCEVAN excelled in clonal breakpoint detection, and Numbat demonstrated high sensitivity for copy number neutral loss of heterozygosity (cnLOH) detection.

Conclusions:

• This benchmark study offers valuable insights into the strengths and weaknesses of current CNA inference tools for scRNA-seq data.
• The findings guide researchers in selecting the most suitable method for their specific cancer genomics applications, improving data interpretation and discovery.