Hello all,
I have several methods and tools like Biobase, Jaspar to predict transcription factors for my genes of interest (in both human and mouse), but they all are giving me too many candidates like +/-40. As I want to test them experimentally, its not possible. Is there a method already available which gives me a handful of potential candidates?
Thanks a lot for your answers.
Regards
Aravind
Hi Aravind, how's things? :-)
What do you mean by "transcription factors FOR your gene of interest"? Are you looking for potential regulators of a particular gene of interest? If so, looking at some ChIP-seq data might be the way to go. TFs that have a peak in the proximity of your gene's TSS are putative regulators. In some / many cases, this might be wrong, but it's still likely to be more accurate than relying on binding motif occurrence alone.
You can look up loads of TF tracks (from ENCODE & others) in the UCSC genome browser (https://genome.ucsc.edu or http://genome.ucsc.edu/ENCODE/dataMatrix/encodeDataMatrixHuman.html ) or alternatively try out GeneProf to search for your gene of interest and it will automatically list TFs binding in the proximity of your favourite gene (search for your gene, then scroll to the bottom of the page for regulators; for mouse & human: http://www.geneprof.org/search.jsp?type=gene&taxon=10090&taxon=9606 ).
Hope that helps!
Cheers,
-Florian.
Hi Aravind,
I'm not an expert on this, but I know people who used MEME suite (http://meme.nbcr.net/meme/) for predicting DNA protein binding sites.
If the number of predicted sites are still too numerous, maybe you can start your functional test by removing them as "group" (if they are close). And after focus where you see a significant impact on the gene transcription.
Best
Guillaume
Perhaps a free trail of genomatix , mathinspector wil give some useful information
https://www.genomatix.de
Hi Aravind, how's things? :-)
What do you mean by "transcription factors FOR your gene of interest"? Are you looking for potential regulators of a particular gene of interest? If so, looking at some ChIP-seq data might be the way to go. TFs that have a peak in the proximity of your gene's TSS are putative regulators. In some / many cases, this might be wrong, but it's still likely to be more accurate than relying on binding motif occurrence alone.
You can look up loads of TF tracks (from ENCODE & others) in the UCSC genome browser (https://genome.ucsc.edu or http://genome.ucsc.edu/ENCODE/dataMatrix/encodeDataMatrixHuman.html ) or alternatively try out GeneProf to search for your gene of interest and it will automatically list TFs binding in the proximity of your favourite gene (search for your gene, then scroll to the bottom of the page for regulators; for mouse & human: http://www.geneprof.org/search.jsp?type=gene&taxon=10090&taxon=9606 ).
Hope that helps!
Cheers,
-Florian.
Hi Aravind,
I think it's normal that you end up with several candidates if you are just scanning sequences around your gene with PWMs. It will give you several false positives for 2 reasons, (1) predicted binding sites which are not functional (2) predicted binding sites which are not functional in your cell type / tissue / condition but that are function somewhere else.
I wouldn't find that weird if a gene would be regulated by 40 TFs under different conditions or cell types, now the question is may be more "how to filter your candidates to select the best ones in your studied system?"
For that I agree with Florian, you can check ChIP data from encode (also DNAseI data to select Tfs for which the binding sites are located in open chromatin region). Here if the cells line/tissues is corresponding to yours it's perfect; if not, you are in case number 2 at least you now that this sites can be functional.
Another thing you can do is to use RNAseq data if available for you tissue of interest to discard TF that are not expressed.
AlsoI think that predictions are more reliable if you start from a set of co-expressed gene rather that from a single gene, because then you can ask for regulators common to you set of genes which will remove noise, but may be it's not something that you can have.
Good luck with your gene!
Delphine
You could also try CentriMo, which has a number of modes for testing for motif enrichment in given localities.
Article Differential motif enrichment analysis of paired ChIP-seq experiments
I'm assuming you have an enhancer of interest in-hand. Phylogenetic conservation is also commonly used as a proxy for "relevance", with the logic that if a small sequence in an enhancer has remained over millions of years of evolution, it's more likely to be important. Predicted binding sites lying within conserved regions might be a good place to start.
If for your biological context same ChIP-Seq / ChIP-chip experiments are available you can also scan your gene promoter for real binding events.
For example you can try this tools: http://amp.pharm.mssm.edu/chea
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools