Skip to main content

Table 6 PSE-HMM is compared to other tools in detecting genome-wide deletions and tandem duplications of size 1 kb, 3 kb, and 5 kb

From: PSE-HMM: genome-wide CNV detection from NGS data using an HMM with Position-Specific Emission probabilities

 

CNV length

1 kb

3 kb

5 kb

  

Precision mean/std

Recall mean/std

F-measure

Precision mean/std

Recall mean/std

F-measure

Precision mean/std

Recall mean/std

F-measure

Duplications

PSE-HMM

0.88/0.04

0.89/0.03

0.89

0.90/0.03

0.97/0.03

0.93

0.88/0.03

0.98/0.01

0.93

m-HMM

1.00/0.00

0.71/0.1

0.83

1.00/0.00

0.76/0.04

0.86

1.00/0.00

0.75/0.03

0.86

Pindel

1.00/0.00

0.95/0.06

0.98

1.00/0.00

0.82/0.11

0.90

1.00/0.00

0.87/0.1

0.93

CNV-seq

0.99/0.01

0.53/0.06

0.69

0.99/0.00

0.53/0.02

0.69

0.99/0.00

0.59/0.09

0.74

Delly

1.00/0.00

1.00/0.00

1.00

1.00/0.00

1.00/0.00

1.00

1.00/0.00

1.00/0.00

1.00

Deletions

PSE-HMM (heterozygous)

0.68/0.10

1.00/0.01

0.81

0.35/0.10

1.00/0.00

0.52

0.70/0.08

1.00/0.03

0.82

PSE-HMM (homozygous)

0.70/0.11

0.89/0.05

0.78

0.77/0.03

0.80/0.00

0.78

0.75/0.03

1.00/0.00

0.86

PSE-HMM (hetero + homo)a

0.69/0.02

0.96/0.02

0.80

0.64/0.02

1.00/0.02

0.78

0.72/0.02

1.00/0.02

0.84

m-HMM (heterozygous)

0.97/0.03

0.50/0.31

0.66

0.99/0.01

1.00/0.00

0.99

0.99/0.01

1.00/0.01

0.99

m-HMM (homozygous)

0.99/0.01

1.00/0.00

0.99

0.99/0.01

1.00/0.00

0.99

0.99/0.01

0.60/0.00

0.75

m-HMM (hetero + homo)a

0.98/0.01

0.67/0.03

0.79

0.99/0.01

1.00/0.02

0.99

0.99/0.01

0.78/0.01

0.87

Pindel

0.98/0.02

0.39/0.08

0.56

0.85/0.10

0.46/0.08

0.59

0.87/0.11

0.45/0.13

0.59

CNV-seq

0.98/0.00

0.92/0.04

0.95

0.98/0.00

0.96/0.01

0.97

0.98/0.00

0.96/0.02

0.97

Delly

0.99/0.01

0.47/0.11

0.64

0.99/0.00

0.51/0.04

0.68

0.99/0.01

0.45/0.12

0.62

  1. The average and standard deviation of the precision (recall) values are calculated based on five different repeats of the whole simulation study with 10× sequencing coverage
  2. a hetero + homo stands for copy loss