Fast index based algorithms and software for matching position specific scoring matrices

BMC Bioinformatics

Table 2 Results of Experiments 1–5.

Experiment 1: 4,034 PSSMs in 20.2 MB protein sequences
p-value	DN00 (total time)	DN00 (search)	LAsearch	ESAsearch +41 sec.
10^-10	65,808	64,939	39,839	41,813
10^-20	38,773	37,706	23,786	24,378
10^-30	21,449	20,362	14,111	13,084
10^-40	9,606	8,533	8,067	5,374
Experiment 2: 220 PSSMs in 37.2 MB DNA
MSS	MatInspector		LAsearch	ESAsearch +32 sec.
0.80	12,773		3,605	202
0.85	12,567		3,189	108
0.90	12,487		2,818	53
0.95	12,445		2,356	12
1.00	12,429		885	1
Experiment 3: 11,411 PSSMs in 4.3 MB protein sequences
E-value	FingerPrintScan		LAsearch	ESAsearch +10.2 sec.
10^-10	4,733		3,423	1.244
10^-20	4,710		486	52
10^-30	4,706		27	10
Experiment 4: 576 PSSMs in 162.9 MB DNA
MSS			LAsearch	ESAsearch +586 sec.
0.85			18,446	318
0.90			16,376	150
0.95			13,764	50
1.00			5,294	1
Experiment 5: 28,337 PSSMs in 4.3 MB protein sequences
raw-th	Blimps		LAsearch	ESAsearch +10.2 sec.
945	271:30:16		16:03:12	11:35:58

Experiment 1: Running times in seconds of the different PSSM searching methods at different levels of stringency, when searching for 4,034 amino acid PSSMs in 59,021 sequences (21.2 MB) from SwissProt. These are the same PSSMs and sequences used in the experiments of [13]. Experiment 2: Running times in seconds of MatInspector, LAsearch, and ESAsearch, when searching 220 PSSMs on both strands of 37.2 MB DNA sequence data at different matrix similarity score (MSS) cutoffs. Experiment 3: Running times in seconds of FingerPrintScan, LAsearch, and ESAsearch when searching all 11,411 PSSMs from the PRINTS database in the RCSB protein data bank (PDB) for different E-values. Experiment 4: Running times in seconds of LAsearch and ESAsearch when searching 576 PSSMs in H. sapiens chr. 6 at different matrix similarity score (MSS) cutoffs. Experiment 5: Running times in hh:mm:ss of Blimps, LAsearch, and ESAsearch when searching all 28,337 PSSMs from the BLOCKS database in PDB. We used a raw score threshold of 945 as suggested in the Blimps documentation for searching large databases. For each experiment, the additional time needed for the construction of the enhanced suffix array is shown in the head of the ESAsearch column.

ISSN: 1471-2105