MetaboMiner – semi-automated identification of metabolites from 2D NMR spectra of complex biofluids

BMC Bioinformatics

Table 1 Performance evaluation using HSQC data collected at pH ~7.2.

Method	Sample	# Cmpds	#Correctly Identified (TP)	FN	FP	Precision (%)	Recall (%)	F score
MetaboMiner-sp	A	27	21	6	6	77.8	77.8	77.8
MetaboMiner-all	A	27	15	12	8	65.2	55.6	60.0
HMDB	A	27	8	19	19	29.7	29.7	29.7
MMCD	A	27	8	19	6	57.1	29.7	39.1
BMRB	A	27	6	21	21	22.2	22.2	22.2
PRIMe	A	27	14	13	6	70.0	51.9	59.6
MetaboMiner-sp	B	21	16	5	5	76.2	76.2	76.2
MetaboMiner-all	B	21	9	12	0	100	42.9	60.0
HMDB	B	21	1	20	20	4.8	4.8	4.8
MMCD	B	21	0	21	0	0	0	0
BMRB	B	21	1	20	20	4.8	4.8	4.8
PRIMe	B	21	6	15	4	60.0	28.6	38.7
MetaboMiner-sp	C	24	22	2	2	91.7	91.7	91.7
MetaboMiner-all	C	24	16	8	3	84.2	66.7	74.4
HMDB	C	24	9	15	15	37.5	37.5	37.5
MMCD	C	24	2	22	0	100	7.4	13.8
BMRB	C	24	3	21	21	12.5	12.5	12.5
PRIMe	C	24	8	16	5	61.5	33.3	43.2
MetaboMiner-sp	D	35	29	6	6	82.9	82.9	82.9
MetaboMiner-all	D	35	16	19	7	69.6	45.7	55.2
HMDB	D	35	9	26	26	25.7	25.7	25.7
MMCD	D	35	4	31	3	57.1	11.4	19.0
BMRB	D	35	7	28	28	20.0	20.0	20.0
PRIMe	D	35	14	21	5	73.7	40.0	51.8

Samples A, B, and C are synthetic cocktail mixtures and sample D is a plasma sample.
(Annotation: MetaboMiner-sp = searched using the biofluid-specific library; MetaboMiner-all = searched using the entire spectral library; TP = true positives; FN = false negatives; FP = false positives).

ISSN: 1471-2105