Single Cell Genotyper (SCG)

Single-cell DNA sequencing has emerged as a powerful tool for studying genetic heterogeneity and clonal evolution in human cancers. However, the analysis of single-cell sequencing data is challenging due to missing values, biased allelic counts, and false genotype measurements from multiple-cell sequencing. To address these issues, the Single Cell Genotyper (SCG) has been developed as an open-source software tool that can robustly infer clonal genotypes from single-cell sequencing data.

SCG uses a statistical model based on a Bayesian hierarchical framework to model the allelic counts of each variant in each cell. The model accounts for the technical noise and biological variability inherent in single-cell sequencing data and estimates the underlying genotypes and their frequencies across the cells. The estimation uses a mean-field variational inference method, which is computationally efficient and can handle large datasets.

The output of SCG includes the inferred genotypes and their frequencies, as well as the posterior probabilities of the genotypes for each variant in each cell. The tool also provides summary statistics, such as the clonal diversity and the clonal composition of the sample, which can be used to compare different samples or conditions. SCG is compatible with various single-cell sequencing platforms and can be customized to fit specific experimental designs.