Skip to main content

Table 1 Program SNV Detection Results for (a) ClinSeq samples and (b) 1000 Genomes samples

From: Evaluation of variant detection software for pooled next-generation sequence data

a
4 PooledSamples %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD
100 % Sample Covg 50 % Sample Covg 25 % Sample Covg 12.5 % Sample Covg
CRISP 99.2 7.8 95.7 98.9 97.3 7.3 95 94.2 88.9 6.5 91.2 76.5 71.3 6.2 82.5 44.9
SNVer 81.9 3.8 89 72.4 74.9 3 85.9 59.1 62.7 2.4 80.1 37.3 48 1.7 73.2 16.6
LoFreq 97.3 8.3 94.5 95.5 93 6.7 93.1 84.1 84 5.2 89.4 63.4 69 4.1 82.5 39.1
VarScan 46.7 0.1 73.3 4.7 47.7 0.1 73.8 6.2 48.9 0.1 74.4 8.1 45 0.3 72.3 6.6
GATK 99.7 6.9 96.4 99.4 98.7 7.4 95.7 96.7 94.7 8 93.3 86.3 85.7 8.7 88.5 65.1
8 Pooled Samples %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD
  100 % Sample Covg 50 % Sample Covg 25 % Sample Covg 12.5 % Sample Covg
CRISP 99.3 7.8 95.8 98.9 97.2 7.4 94.9 94.1 88.9 6.8 91.1 77.5 71.2 6.7 82.2 46.1
SNVer 79.9 3.6 88.1 65.7 69.4 3.2 83.1 47.1 55.5 2.8 76.3 25.3 42.5 2.4 70 9.9
LoFreq 96.7 7.3 94.7 93.1 91.8 6.5 92.7 79 82.8 5.4 88.7 56 70.7 4.3 83.2 31.8
VarScan 28.8 0.1 64.4 0 29.2 0.1 64.5 0.1 29.8 0.1 64.8 0.1 30.4 0.2 65.1 0.3
GATK 98.5 8.6 94.9 96.4 98 8.5 94.7 95.1 94 10.1 91.9 86 83.9 11.4 86.3 64
16 Pooled Samples %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD
  100 % Sample Covg 50 % Sample Covg 25 % Sample Covg 12.5 % Sample Covg
CRISP 99.1 7.7 95.7 98.5 96.7 7.6 94.6 93.3 87.7 7 90.4 76.6 69.3 7 81.2 46.3
SNVer 66.9 3.5 81.7 42.9 53.7 3.4 75.1 23.8 42.4 3.2 69.6 10.8 33.2 3 65.1 3.6
LoFreq 94.9 6.4 94.2 87.6 88.5 6 91.3 69.8 78.5 5.5 86.5 44.7 67 4.8 81.1 22.5
VarScan 18.1 0.1 59 0 18.2 0.1 59 0 18.4 0.1 59.1 0 18.7 0.1 59.3 0
GATK NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA
32 Pooled Samples %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD
  100 % Sample Covg 50 % Sample Covg 25 % Sample Covg 12.5 % Sample Covg
CRISP 98.7 7.9 95.4 97.5 95.8 7.6 94.1 91.8 86 7.1 89.5 74.4 67.5 7.3 80.1 46.2
SNVer 41.8 4.2 68.8 11 34.6 4.2 65.2 5 29.3 3.8 62.7 2.3 24.5 3.8 60.4 0.6
LoFreq 90.7 5.4 92.7 77.5 82.2 5.4 88.4 55.7 71.1 5.3 82.9 31.2 60.5 5.2 77.6 13.8
VarScan 11.4 0 55.7 0 11.5 0.2 55.7 0 11.5 0.2 55.7 0 11.6 0.2 55.7 0
GATK NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA
b
4 Pooled Samples %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD
  100 % Sample Covg 50 % Sample Covg 25 % Sample Covg 12.5 % Sample Covg
CRISP 99.2 4.4 97.4 98.5 90.9 4 93.5 80.3 74.2 3.7 85.3 48.7 55.5 3.3 76.1 22.4
SNVer 86.5 1.3 92.6 74.8 70.9 0.9 85 47.3 50.3 0.4 74.9 18.5 33.2 0.6 66.3 4.4
LoFreq 96.8 0.4 98.2 94.1 87.3 0.2 93.5 73.9 71 0.1 85.5 44.3 48.9 0 74.4 19.7
VarScan 44.6 0 72.3 0.9 45.1 0 72.5 3.2 42.3 0 71.2 3.8 33.4 0 66.7 1.4
GATK 99.9 0.3 99.8 99.8 97.5 0.2 98.6 93.8 89.3 0.2 94.6 74.9 73.9 0.3 86.8 44.7
8 Pooled Samples %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD
  100 % Sample Covg 50 % Sample Covg 25 % Sample Covg 12.5 % Sample Covg
CRISP 99.2 4.3 97.5 98.4 91.3 4.1 93.6 81.4 75.5 3.7 85.9 50.4 57.7 3.2 77.2 23.8
SNVer 80.5 2.1 89.2 62 61.8 1.8 80 30.1 44.5 0.8 71.8 9.3 30.6 0.8 64.9 1.6
LoFreq 95.4 0.4 97.5 89.7 84.4 0.2 92.1 64 68.5 0.2 84.1 33.3 52.5 0 76.3 13.1
VarScan 25.5 0 62.7 0 26 0 63 0 26.9 0 63.4 0 25.7 0 62.8 0.1
GATK 99.6 1.5 99.1 99.1 97 0.7 98.1 92.6 88.4 0.5 93.9 72.9 72.9 0.4 86.2 42.7
16 Pooled Samples %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD %Sen %FP %BA %SD
  100 % Sample Covg 50 % Sample Covg 25 % Sample Covg 12.5 % Sample Covg
CRISP 99.1 4.2 97.4 98.1 91.2 4 93.6 81.5 74.2 3.5 85.3 48.3 57.3 3.3 77 24.5
SNVer 61.3 4.4 78.4 27.8 47.6 3.3 72.1 9.8 36.1 1.6 67.3 1.7 27.3 0.9 63.2 0.3
LoFreq 91.8 0.6 95.6 80.8 78.9 0.2 89.3 50.3 63.2 0.1 81.5 22.4 49.2 0.2 74.5 7.4
VarScan 14.9 0 57.4 0 15.1 0 57.6 0 15.3 0 57.6 0 15.6 0 57.8 0
GATK NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA
  1. GATK was unable to process the 16 or 32 pooled sample pools (see runtime results). Pools were run in groups of 8 for the ClinSeq samples and groups of 4 for the 1000 Genomes samples, except for LoFreq runs, which ran on individual pools, before grouping the results in sets of 8 (ClinSeq) or 4 (1000 Genomes) to calculating sensitivity, false positive rate, balanced accuracy, and singleton detection rate. Numbers reported in bold face represent the better performance values for each column
\