Differential performance of strategies for single-cell whole-genome amplification

Summary: Single-cell genomics enables studying tissues and organisms at the highest resolution. However, since a cell contains a small amount of DNA, single-cell DNA sequencing (scDNA-seq) typically requires single-cell whole-genome amplification (scWGA). Unfortunately, scWGA methods introduce technical biases that complicate the interpretation of scDNA-seq data. We compared six scWGA methods, three MDA (multiple displacement amplification; GenomiPhi, REPLI-g, and TruePrime) and three non-MDA (Ampli1, MALBAC, and PicoPLEX), on 206 tumoral and 24 healthy human cells. scWGA methods performed differently depending on the parameter of interest. REPLI-g minimized regional amplification bias, while non-MDA methods showed a more uniform and reproducible amplification. Ampli1 exhibited the lowest allelic imbalance and dropout, the most accurate insertion or deletion (indel) and copy-number detection, and a low polymerase error rate. However, REPLI-g yielded higher DNA quantities, longer amplicons, and greater genome coverage. We offer a comprehensive guide for selecting a scWGA approach, outlining trade-offs that influence the interpretation of scDNA-seq data. Motivation: Single-cell whole-genome amplification (scWGA) is a critical step for the genomic study of single cells, yet its impact on downstream analysis remains poorly understood. Previous studies were not comprehensive in the number of methods tested and the metrics assessed. Besides, they have mostly been conducted or replicated by the original authors of a method rather than by independent third-party laboratories. Here, we test six commercially available scWGA methods to understand their performance and propose a guide for selecting the best scWGA approach for studies on single-cell genomics.