Hi there,
I'm performing some (very basic) bioinformatic analysis of our gene of interest, part of which is looking at the different transcription factors that may regulate it using the ENCODE Regulation track on the UCSC Genome Browser.
One particular question I'm interested in asking is: what promoters in human cells bind the same set of transcription factors as my GoI? In more specific language, which genes in the genome have a strong ChIP signal for two (or more) defined transcription factors within, say, 3kb upstream of their transcriptional start site.
I'm very new to this kind of whole-genome dataset analysis and as a result am a near-novice; any advice anyone could give on how to do this kind of analysis in a reasonable time would be greatly appreciated!
I'm expecting to need to download various genome viewer(s) and or datasets etc., but at largely at sea regarding: a) what best to use for this and b) the method/approach or coding required to e.gh. define the TSS distance and the requirement for double-positive ChIP signals is beyond me.
Alternatively, if there is a useful programme/took/database for this that I am unaware of, please do let me know!
Thanks in advance for all help!
It wasn't clear from your question whether you know the transcription factors that are acting on your GoI (and related), or if this is an open question. If you know the TFs that you are interested, you might want to start with files containing predicted peaks (often in BED format) for your TFs, and see where they intersect (BEDtools is a common tool for this). You could then see where your intersected TF binding regions overlap the 3kb upstream of a gene.
If you are trying to discover any combination of TFs that are binding near your gene, it's a more complicated problem, one that I think that can be solved by hierarchical clustering (see what other genes cluster with your GoI).
Great, thanks Joseph! To be clear, I do know which TFs I'm keen to look at - but just wanted to try a couple of different pairwise combinations.
Just to clarify, is BEDTools a standalone programme or an extension to something like the UCSC Genome Browser? And is there a particular way of defining the, say, 3kb upstream? Or is this a built-in option to the programme? Apologies for the many questions - very much a "learning by doing" approach to bioinformatics!
Regards,
Dan
Bedtools is a command-line suite of tools (in a single program). http://bedtools.readthedocs.org/en/latest/
Depending on your comfort with Unix-type "make" command, you can install it locally. Alternatively, your institution may have a shared computing resource (e.g. a research computing cluster) that may have it pre-installed. Thirdly, you can use the tools at www.usegalaxy.org. To start, I'd probably recommend the third option, since their video tutorials are pretty fantastic.
Depending on your organism, you can find pre-computed coordinates for all promoters (usually defined as the region immediately adjacent to predicted transcription start sites). Relatively simple command-line tools can adjust the width of the interval, for example, using Awk. For example, from an annotation file (GFF or BED), you would just start at the first nucleotide of each gene, subtract 3000 bases if the gene is forward orientation, or add 3000 if the gene is reverse orientation. Now that I think about it, you could easily do this in Excel.
If you are going to be doing a fair amount of bioinformatics in the future, you might want to start learning biopython or Bioconductor (in R). I just completed the coursera "Bioconductor for Genomic Data Science" course (the free, non-signature track version), and there are ways to do everything that you described, in Bioconductor. The problem is, you first need to learn lots about Bioconductor...
I hope this is helpful!
Hi Joseph,
Thanks for going into so much detail - it's really useful to have someone put everything together in clear, plain-language format! I'll definitely check out usegalaxy, and the Coursera course sounds like a great idea. I'm not too confident about Unix - yet - but I feel it's time that I bit the bullet and learnt it.
Thanks again for all your help! I'm looking forward to giving it all a try.
Dan
Most transcription factors exhibit tissue or cell-type specific binding. Unless you are studying the same cell lines (or closely similar) to those used in the ENCODE studies, those reference datasets might have limited utility for your experimental question/system.
I agree with Joseph that Galaxy is a easy and powerful tool for beginners, but one caveat is that large datasets can get truncated transferring between UCSC and Galaxy. You will probably do better downloading the TFBS track (wgEncodeRegTfbsClusteredV3.bed.gz) directly at the following link http://hgdownload.cse.ucsc.edu/goldenPath/hg19/encodeDCC/wgEncodeRegTfbsClustered/
You can then import into Galaxy. In terms of command-line tools, I'd recommend BEDOPS toolkit compared to BEDTOOLs, based on both its speed and easy-to-understand documentation.
http://bedops.readthedocs.org/en/latest/
Bottom line is that, yes, this analysis is straightforward if you know which TFs you want to examine. Whether the ENCODE cistromes are relevant for your cell/tissue type of interest is for you to decide.
Hi Daniel,
Thanks for your comment - it's a very good point about the cell-line specificity, and I'm definitely planning to: (a) look at a couple of different cell line data sets; and (b) align them as closely as possible to the cell type I'm currently using.
I'll definitely check out the BEDOPS kit and pull down the TF datasets as you suggest - thanks for your suggestions!
Regards,
Dan
Hi Joseph and Daniel,
I just wanted to say that, with your help, I got exactly the answer I wanted using the UCSC dataset and UseGalaxy as a GUI - I've now been using it to answer other (very simple) bioinformatics problems (defining intra-intronic TF sites etc.).
Thanks very much for all your help!
All the best,
Dan
Aw, that's fantastic, very nice of you to say, and glad to hear it worked out! Have fun!
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts