A comparative evaluation of sequence classification programs

BMC Bioinformatics

Table 6 Performance of composition-based programs

Program	PhyloPythia 961 bp	PhymmBL 243 bp	RAIphy 238 bp	Mean	Training
Percentage of sequence classified
NBC	100	100	100	100
PhyloPythiaS	3.5	3.1	3.3	3.3
PhymmBL	100	99.7	100	99.9
RAIphy	100	100	100	100
Sensitivity (percentage)
NBC	95.4	97.5	99.4	97.4
PhyloPythiaS	3.1	1.8	2.2	2.4
PhymmBL	48.4	96.8	81.9	75.7
RAIphy	54.8	31.8	48.0	44.9
Precision (percentage)
NBC	95.4	97.5	99.4	97.4
PhyloPythiaS	88.1	58.5	66.1	70.9
PhymmBL	48.4	97.0	81.9	75.8
RAIphy	54.8	31.8	48.0	44.9
CPU Runtime (minutes)
NBC¹	13496	3595	17573	11555	1217
PhyloPythiaS²	297	180	506	328	4320
PhymmBL¹	15600	1035	23508	13381	2880
RAIphy³	105	25	122	84	30
Memory Usage (Megabytes of RAM)
NBC	200	200	200	200
PhyloPythiaS⁴	100	100	100		100
PhymmBL⁴	100	100	100	100
RAIphy	500	335	400	412

Measurements of sensitivity, precision, and resource consumption on three simulated data sets.
¹analysis performed on an AMD Opteron 250 (2.4 GHz) Sun Fire V40z with 32 GB RAM.
²analysis perfomed on an AMD Opteron 248 (2.2 GHz) workstation with 8 GB RAM.
³analysis performed on a 2.66 GHz Intel Core i7 MacBook Pro running Mac OS X 10.7.1 with 8 GB 1067 MHz DDR3 RAM.
⁴input sequences were broken up into smaller files.

ISSN: 1471-2105