Skip to main content

Table 4 Perfomance of the individual filters evaluated to discriminate between true and false variants by the AUC values from ROC curve, grouped by type of variants (SNV or Indel) and status of the genotype call (homozygote or heterozygote) according to the depth of sequencing (LC or HC dataset)

From: GATK hard filtering: tunable parameters to improve variant calling for next generation sequencing targeted gene panel data

  SNV Indel
  Homo Het Homo Het
HC dataset
BQRS 0.73 0.53 0.5 0.53
RPRS 0.57 0.61 0.52 0.68
CRS 0.5 0.51 0.53 0.5
DP 0.79 0.8 0.76 0.6
MQ 0.55 0.5 0.6 0.63
MQRS 0.52 0.53 0.58 0.58
GQ 0.65 0.95 0.77 0.77
ADT 0.96 0.8 0.56 0.94
ADTL 0.8 0.77 0.72 0.94
FS 0.51 0.62 0.51 0.54
LC dataset
BQRS 0.58 0.65 0.5 0.52
RPRS 0.53 0.5 0.54 0.74
CRS 0.52 0.5 0.51 0.51
DP 0.63 0.67 0.52 0.62
MQ 0.51 0.54 0.54 0.52
MQRS 0.52 0.52 0.51 0.5
GQ 0.79 0.99 0.53 0.97
ADT 0.98 0.99 0.5 0.92
ADTL 0.67 0.98 0.54 0.98
FS 0.5 0.54 0.5 0.54