Skip to main content
Figure 1 | BMC Bioinformatics

Figure 1

From: VarBin, a novel method for classifying true and false positive variants in NGS data

Figure 1

Read coverage depth effect on likelihood scores. Each data point represents one variant change per one sample from the study family's genomes and 38 background data files. Blue data points passed the GATK best practice variant filters and are called variants in the vcf file. Red data points are "non-variants" which did not pass these same filters. Homozygous variants were given an × axis value of zero to separate them from the heterozygous variant distributions. Values beyond the plot axis limits are shown at the plot edges. A and B) Scatter plot of the Phred-scaled likelihood ratio (PLR) versus coverage depth for two variant sets. A) Variant data set enriched for true variants. For this plot, variants in the study family proband within chromosome 1 were limited to between 10 and 20% allele frequency in 1000 Genomes data (20,000 variants total). B) Variant data set enriched for false positive variants. For this plot variants in the study family proband at chromosomes 1 - 22, then limited to variants not found in the 1000 Genomes data or either parent (14,500 variants total). C and D) Scatter plot of the Phred-scaled likelihood ratio divided by coverage depth (PLRD) versus coverage depth for two variant sets. C) Data from panel A divided by coverage depth. D) Data from panel B divided by coverage depth

Back to article page