Skip to main content

Table 2 SNP calling precision/recall test on data from human chromosome 20, compared to a gold standard coming from the “1000 genomes project”

From: Reference-free compression of high throughput sequencing data with a probabilistic de Bruijn graph

HG00096 chrom 20

Prog

Precision

Recall

Compression ratio

lossless

85.02

67.02

2.95

SCALCE

85.15

66.13

4.1

FASTQZ

85.46

66.63

5.4

LIBCSAM

84.85

67.09

8.4

FQZCOMP

85.09

66.61

8.9

LEON

85.63

67.17

11.4

RQS

85.59

67.15

12.4

no quality

57.73

68.66

-

  1. No quality means all qualities were discarded and replaced by ’H’. The ratio is given by the original quality size divided by the compressed size. For the lossless line, the best compression ratio obtained by lossless compression tools is given (obtained here with FQZCOMP). Results are ordered by increasing compression ratio
  2. Best overall results are in bold