Latent Semantic Indexing of PubMed abstracts for identification of transcription factor candidates from microarray derived gene sets

Table 6 Comparison of TF rankings produced by four different methods for the five datasets.

Dataset	TF knockout	tf-normal	log-entropy	co-occurrence	CORE_TF
Interferon	Rel	22	3	90	[1--86]^*
IL-4	Stat6	2	16	17	69
Tunicamycin	Ddit3	21	128	1	-^†
GH	Stat5a/b	13	50	29	9
TGF-β1	Nfic	241	233	[205--433]^§	289

^*Using CORE_TF, Rel could be ranked anywhere between 1 and 86 as its associated motif V$CREL_01 had a p-value = 0 (rank 1) along with 85 other motifs. ^†The motif for Ddit3, V$CHOP_01 was not ranked by CORE_TF. ^§Using the abstract co-occurrence model, the TF Nfic could be ranked anywhere between 205 and 433 as it did not share any abstracts with any gene in the TGF-β1 dataset and therefore had a p-value of 1 along with 228 other TFs.

ISSN: 1471-2105