Skip to main content

Table 6 Results by the application of selection parameters and their thresholds on simulated datasets

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

  TV
N (%)
FV
N (%)
Variant selected by hard filtering %
HC dataset
 Homo SNVs
   Overall 2,382 (66.6) 1,195 (33.4) 93.9
   Selected 2,238 (98.6) 31 (1.3)  
 Het SNVs
   Overall 87,361 (99.8) 177 (0.2) 98.6
   Selected 86,166 (99.9) 24 (0.03)  
 Homo indels
   Overall 54 (0.12) 43,871 (99.8) 27.7
   Selected 15 (75) 5 (25)  
 Het indels
   Overall 17.305 (45.8) 20,410 (54.1) 84.6
   Selected 14,646 (94) 935 (6)  
LC dataset
 Homo SNVs
   Overall 2,084 (12.38) 14,721 (87.6) 96.9
   Selected 2,020 (92.2) 171 (7.8)  
 Het SNVs
   Overall 95,119 (97.6) 2,322 (2.3) 99.4
   Selected 94,602 (99.9) 80 (0.08)  
 Homo indels
   Overall 154 (0.4) 45623 (99.6) 100
   Selected 154 (0.4) 45623 (99.6)  
 Het indels
   Overall 7,502 (17.6) 34,889 (82.3) 43
   Selected 3,226 (99.1) 27 (0.8)  
  1. % have to be intended as the percentage of unfiltered variants for “overall”calls and as the percentage of alterations which were not filtered out in the hard filtering process for “selected”calls; % of selection indicates the amount of variants selected from the total callset. TV true variants, FV false variants