Skip to main content

Table 3 Comparison of MaxSSmap and MaxSSmap_fast to a GPU and a SIMD high performance Smith-Waterman implementation

From: MaxSSmap: a GPU program for mapping divergent short reads to genomes with the maximum scoring subsequence

 (a) Percent of 100,000 251 bp reads mapped correctly to the E.coli genome. Shown in parenthesis are incorrectly mapped reads and remaining are rejected

Div

MaxSSmap_fast

MaxSSmap

CUDASW++

SSW

Reads without gaps

 

.1

95 (0.4)

96 (0.4)

94 (0.9)

97 (3)

.2

95 (0.6)

95.3 (0.6)

94 (1)

97 (3)

.3

90 (1.1)

94.2 (0.9)

93 (1.3)

96 (4)

Reads with gaps

 

.1

92 (1.5)

93.1 (1.9)

94 (0.9)

97 (3)

.2

90 (1.7)

92.5 (2.1)

92 (1)

96 (4)

.3

81 (2.8)

89.9 (3.5)

92 (1.4)

95 (5)

(b) Time in minutes to map 100,000 251 bp reads to the E.coli genome

Div

MaxSSmap_fast

MaxSSmap

CUDASW++

SSW

Reads without gaps

 

.1

20

28

164

1288

.2

20

28

164

1275

.3

20

28

164

1255

Reads with gaps

 

.1

20

28

163

1283

.2

20

28

162

1266

.3

20

28

162

1235

  1. These are simulated Illumina reads and contain realistic base qualities generated from Illumina short reads. Each divergence represents the average percent of mismatches in the reads. So 0.1 means 10% mismatches on the average. The gaps are randomly chosen to occur in the read or the genome and are of length at most 30.