Uncovering transcriptional interactions via an adaptive fuzzy logic approach

BMC Bioinformatics

Table 2 The performances of AdaFuzzy and the other approaches, checked against validated TIs of the 19 TFs from YEASTRACT.

Methods	True positives	False positives	False negatives	True negatives	Sensitivity	Specificity	False Positive rate	False Negative rate
AdaFuzzy	6736	188	2495	108942	73.0%	99.8%	0.2%	27.0%
ChIP*	2799	108	6432	109022	30.7%	99.9%	0.1%	69.3%
cPSMF	1729	0	7502	109130	18.7%	100.0%	0.0%	81.3%
GRAM	1615	0	7616	109130	17.5%	100.0%	0.0%	82.5%
COGRIM	1406	0	7825	109130	15.2%	100.0%	0.0%	84.8%
ReMoDis.	608	0	8623	109130	6.6%	100.0%	0.0%	93.4%

2907 TIs identified by 'ChIP' were obtained by ChIP-chip experiment presented in Harbison et al. (2004) using criterion p-value < 0.001.
^† False-positives are assumed zero to estimate the best (upper bound of the) performance yielded by the other competitive methods.

ISSN: 1471-2105