Skip to main content
Fig. 1 | BMC Genomics

Fig. 1

From: Systematic evaluation of signal-to-noise ratio in variant detection from single cell genome multiple displacement amplification and exome sequencing

Fig. 1

Loci dropouts, allelic mismatch and read depth variation from sparse cell assays. The correlation between molecular marker dropout and low cell input is evident from microsatellite genotyping (STR) based on 21 different loci compared to bulk DNA from OCI-AML3 (a). This dropout was manifested by a complete absence of a specific marker loci (A, red), unambiguous loss of heterozygosity (a, orange) or partial dropout (a, blue), represented as imbalance in heterozygosity. Intersection of variants from replicate genome amplification and exome sequencing (b) shows allelic mismatch between duplicate assays, i.e. degree of reproducibility, as an approximately exponential function of cell input. Inclusion of all raw GATK-passed variants led to an allelic mismatch of approximately 50% (b, blue line). The proper inclusion of only coding variants decreased this number to one-third (B, orange), whereas setting a coverage threshold (red) did not additionally improve the result. Allelic read depths are highly influenced by the number of cells used for amplification as demonstrated here by the relative comparison of phred-scaled likelihood of each being called heterozygous (P = 10(-PL/10)) to one replicate of the 50-cell assays (c, 0.51 < ρSpearman < 0.84) – directly a result of read depth variation of both reference and alternate alleles. Comparison of the single nucleotide variant (SNV) sets from 1 to 50 cells against the reported somatic SNVs in the COSMIC database (286 SNVs, Cell Lines Project v85, OCI-AML3) revealed only a minor decrease in variant overlap (d, blue) as cell input decreased. This was also the case, when focusing on the more confident subset of 31 SNVs marked as reported previously in COSMIC (d, orange)

Back to article page