I've done a TFBS analysis with the Jaspar Core dataset of plants on a plant promotor in reverse orientation (start with GTA in complement strand ). It seems that the TFs that are detected  in 3' to 5' orientation and not in the 5' to 3' orientation of the sequence, as one would expect. Also the core sequence within the matrices do not seem to be respected during detection, e.g. in sequence TAAATGACGTCTTT the following sequence is annotated ATGACGT, while the core sequence is TGACGTG (bZIP910, last G is not respected, so this could be a false positive).

Hi, Myckel

I've searched your sequence with our realization and with official JASPAR database search engine. Both ways I've got hits on ATGACGT (this, for example, result I've got on jaspar.genereg.net)
http://files.myopera.com/lao/albums/3686792/jaspar.png

Also I don't think, that found subsequence is flase positive, cause bZIP910 matrix detects hits of ATGACGT and CTGACGT sequences (you can look at following Alignment Logo to check it):
http://files.myopera.com/lao/albums/3686792/bZIP910.png

I've checked if detection in complement strand is wrong and found that all is right here too. Can you provide some more information about wrong weight matrix search behavior, so if we really have mistakes in algorithm, we can fix it in the next UGENE release

Hi Oleg,

I now see that the matrix is not the same as in the literature it references (I was only looking at the literature and the result of ugene).

About the complementary strand, it seems I was misunderstanding the way TFs are detected (they can apparently be in both orientations 5'-3' and 3'-5' on the same strand).

Thank you for your effort anyway.

Myckel