Fig 1.
BAM files from BWA, Isaac aligner, and Bowtie2 were paired with each of GATK, Isaac variant caller, and SAMTools run both with and without additional filtering (VQSR, BAQ, and LowGQX respectively). Output vcf files were regularized using custom code and variants from GIAB high quality regions taken forward to generate false positive and false negative rates.
Fig 2.
Software concordance Venn diagrams.
Merged variant calls for each tool were overlapped with other tools of the same variety for each variant type. Aligners are compared in row 1, variant callers without filtering in row 2 and variant callers with filtering in row 3.
Fig 3.
Software concordance ROC curves.
Merged variant calls for each tool were calculated and ROC curves generated using the genome quality score. Aligners are compared in row 1, variant callers without filtering in row 2, and variant callers with filtering in row 3.
Table 1.
Total SNV calls and tool-specific SNV calls for each aligner and variant caller run with and without filtering.
Table 2.
Total deletion calls and tool-specific deletion calls for each aligner and variant caller run with and without filtering.
Table 3.
Total insertion calls and tool-specific insertion calls for each aligner and variant caller run with and without filtering.
Table 4.
Variant calls grouped by frequency of detection.
Table 5.
SNV calls for GIAB data at simulated contamination levels.
Table 6.
Melanoma cell line control variant calls overlapping annotated ClinVar melanoma risk factors.