Methods for Single-Cell DNA Sequencing Analysis with Application to the Human Brain

Abstract

Interest in SMs has recently increased beyond the context of cancer, driven by efforts to understand SMs in pre-cancerous states, degenerative disease and aging. However, there are unique challenges to detecting SMs in normal tissues since they are generally shared by fewer cells than in tumors and, in the case of post-mitotic cells such as neurons, may not be shared at all. These rare SMs are more difficult to detect when analyzing bulk cell samples because they are often not sampled or are discarded as possible artifacts. Single-cell DNA-sequencing (scDNA-seq) offers the ability to detect these rare SMs—even those private to a single cell—but has been impeded by artifacts introduced during the whole-genome amplification (WGA) step required for analysis by modern short-read DNA sequencing platforms. Here we present several methodological advances for detecting somatic single nucleotide mutations and small insertions and deletions (indels) in deep scDNA-seq data. First, we provide a model for a pervasive single-cell WGA artifact termed allelic imbalance and construct a genotyper based on this model (SCAN-SNV). We then take advantage of a new and improved whole-genome amplification chemistry (Primary Template-directed Amplification; PTA) to build a genotyper (SCAN2) that increases sensitivity using the concept of mutation signatures and that can detect somatic indels. Finally, we generate PTA scDNA-seq data for 52 single neurons from 17 human brains aged 0-104 years and characterize the rates and patterns of SM accumulation in the aging human brain.