Monovar: single-nucleotide variant detection in single cells

Abstract

Current variant callers are not suitable for single-cell DNA sequencing, as they do not account for allelic dropout, false-positive errors and coverage nonuniformity. We developed Monovar (https://bitbucket.org/hamimzafar/monovar), a statistical method for detecting and genotyping single-nucleotide variants in single-cell data. Monovar exhibited superior performance over standard algorithms on benchmarks and in identifying driver mutations and delineating clonal substructure in three different human tumor data sets.

Figure 1. Monovar Algorithm and Performance in a Normal Cell Line

(a) Monovar variant detection flowchart. (b)–(e) Evaluation of Monovar, GATK and Samtools for the detection of SNVs in a single cell exome sequencing dataset generated from a normal isogenic fibroblast cell line. (b) Venn diagram showing the number of TPs called by different algorithms. (c) Venn diagram showing the number of FPs called by different algorithms. (d) Comparison of the SNV spectrum for FP errors detected using different variant detection algorithms. (e) Precision vs Detection Efficiency (Recall) curve for Monovar.

Figure 2. Application of Monovar to Human Tumor Samples

Monovar was applied to detect somatic mutations in datasets from three human tumor samples, including a triple-negative breast cancer (a), a muscle-invasive bladder cancer (b) and a childhood acute lymphoblastic leukemia patient (c). Multi-dimensional Scaling analysis (left panels) and hierarchical clustering (right panels) were performed using the single cell genotype matrices to identify subpopulations of single cells that shared common sets of somatic mutations. Mutations in genes that were previously detected in these studies are listed in black, while new mutations identified by Monovar are listed in red.