LIMPIC: a computational method for the separation of protein MALDI-TOF-MS signals from noise

BMC Bioinformatics

Table 2 Performance comparison of peak-detection algorithms.

Protein concentration		2.5 fmol	5 fmol	7.5 fmol	10 fmol	25 fmol	50 fmol
Number of manually detected peak classes		7	7	10	8	10	11
LIMPIC	Number of detected classes	8	7	10	8	10	11
	Number of missed classes	0	0	0	0	0	0
	Number of incorrect classes	1	0	0	0	0	0
APEX	Number of detected classes	27	18	34	24	25	21
	Number of missed classes	1	1	3	1	0	2
	Number of incorrect classes	21	12	27	17	15	12
CENTROID	Number of detected classes	40	52	48	24	25	27
	Number of missed classes	1	1	1	0	0	0
	Number of incorrect classes	34	46	39	16	15	16
CROMWELL	Number of detected classes	14	15	14	16	21	24
	Number of missed classes	3	3	6	2	3	4
	Number of incorrect classes	10	11	10	10	14	17

LIMPIC, APEX, CENTROID and CROMWELL methods were assessed from in-vitro protein spectra at different cytochrome C dilutions, with respect to the number of detected peak classes, false negatives and false positives. Only the peak classes that could be found across replicate spectra with a peak detection rate (PDR) larger than 0.5 were considered for the analysis.

ISSN: 1471-2105