Tf-Target Enrichment Analysis

Question

[Jimbo]

Given a list of DE genes, eg from a microarray experiment, of which 2 are known TFs, how can I see if any of the other genes are known or predicted to be regulated by the known TFs?

Many thanks,

Jim

[Khader Shameer]

Have you checked a discussions at BioStar: How To Identify Targets Of A Transcription Factor ? ?

[Larry_Parnell]

If the answer is useful, please vote them up, Jimbo, and try to select the best for your purposes. This is an essential element of what makes BioStar so beneficial.

[Jimbo]

Ahh yes, apologies! I have selected the best answer for other people, although i would like to give credit to Anima mundi too, because he corresponded with me on further queries, so I "voted him up". I hope this is the correct procedure. Cheers

Answer 1

[Stephane Plaisance]

I used to run these querries at:

• http://dire.dcode.org/
• http://genecodis.dacya.ucm.es/analysis/ (you can use Transcription factors only for this appplication)
• http://atted.jp/ (finds coregulated genes, you can use the output to feed the former tools)

I am sure there are other sites all based on some adapatation of the UCSC conserved TFBS track

to go deeper and across species for a given region, use * http://bioit.dmbr.ugent.be/ConTra/index.php *... google it

these web-tools ar eeasy and give you results you of course will need to validate.

Good luck

Answer 2

[Anima Mundi]

Hello, I would suggest you the PSWM usage. You could download the matrices of the TFs, and use them to predict TFBS among the genes of the list. For known targets I would suggest iHOP, a tool capable of retrieving literature-based information about several genes.

[Jimbo]

This is what I was expecting. Is there any software you know of that can do this, i.e. you input a list, it automatically detects TFs, finds their "PSWM" matrices (or other), then automatically scans the remaining genes to look for TFBS (and perhaps ranks them, respectively for each TF)

[Jimbo]

Oh, and thanks a lot for the answer, of course :)

[Anima Mundi]

For what you desire I would suggest Motifviz's Possum. You can input your list using the JASPAR Core database (a database of PSWMs); the output is a series of stressed TFBSs for some TF in your sequences. This is a "quick and dirty" method, you could need something more complex and reliable, depending on your scopes.

[Anima Mundi]

For what concerns the detection of the TFs you should somehow use the annotation of your sequences, then make a script to retrieve only the desired TFs from the JASPAR database, generate a file containing the resulting list of PSWMs and input it. But it is not useful, I would simply input the whole database.

[Jimbo]

OK I see, so put in everything, then filter the output for the TFs that were in the original list?

[Anima Mundi]

Exactly, I would do so.

[Ying W]

I wanted to do the same thing and some of the replies can be found here. Unfortunately, I haven't gotten around to doing it yet (waiting for data on more TFs)