Finding specific subsets of genes related with known transcription factors

Question

[david.bolumar]

Hi there,

It's the first time I use this platform so I hope it reaches to many people.

I have just got DNA sequencing results from an experiment where I had different conditions. I made comparisons from those conditions and got those Transcription Factor Binding Sites (TFBSs) that where enriched in each of them.

My field of work is reproduction (focusing in embryo development and implantation). I want to extract those genes that are regulated by each TF (associated with the TFBSs) and which are related with functions in embryo development and implantation. I think that may be filtering by GO terms may be the best way to do this. In that case, for each TFs, I would also like to get a parameter showing the contribution of GO terms related to reproduction from the total GO terms for the TF.

I've been reading about a method called PASTAA that may help, but I don't really know how it works or even if it would be useful for my problem.

Thanks in advance!

[tarek.mohamed]

Hi David,

Try Homer, it should be useful

http://homer.ucsd.edu/homer/microarray/go.html

http://homer.ucsd.edu/homer/introduction/programs.html

Tarek

[Friederike]

by TFBS do you mean you have a motif matrix or a the regions in the genome (i.e. the coordinates) where a factor bound?

[BioinfGuru]

You say you have enriched TFBSs - To ensure we can answer your question, and to avoid any confusion with terminology and interpretation, can you tell us please how you got these "TFBSs"? It is more likely that you actually have a set of enriched motifs. If you do have the sites already.... then for each site you only need look for the nearest downstream gene within the same TAD

[david.bolumar]

In order to get the TFBSs sequences we just sequenced the genomic regions within our target samples. Afterwards, we conducted a comparative analysis of the TFBSs regions using the ENSEMBL database as the source.

We actually thought about looking for the genes sourrounding the TFBSs, but this seems quite imprecise as, as you know, some of the TFs present trans-acting activity and may regulate promoters quite far up/downstream of their position.

Thanks again!

Answer 1

[Alex Reynolds]

If you have the TF binding sites in UCSC BED format, you can use BEDOPS convert2bed to convert gene annotations to BED, bedops to make a file of gene promoters (say, a region 500 nt upstream of the gene TSS), and bedmap to associate binding sites with the promoters of genes.

Get some gene annotations and write them to a BED-formatted file:

$ wget -qO- ftp://ftp.sanger.ac.uk/pub/gencode/Gencode_human/release_27/gencode.v27.basic.annotation.gff3.gz \
    | gunzip --stdout - \
    | awk '$3 == "gene"' \
    | convert2bed -i gff - \
    > genes.bed

Or replace this step with whatever annotation source you prefer, for your reference genome.

Make a file of promoter regions from the genes:

$ awk -v OFS="\t" '($6 == "+") { print $1, $2, ($2+1), $4; }' genes.bed | bedops --range -500:0 --everything - > promoters.for.bed
$ awk -v OFS="\t" '($6 == "-") { print $1, ($3 - 1), $3, $4; }' genes.bed | bedops --range 0:500 --everything - > promoters.rev.bed
$ bedops --everything promoters.for.bed promoters.rev.bed > promoters.bed

Sort your BED-formatted TF binding sites:

$ sort-bed tfbs.unsorted.bed > tfbs.bed

Map binding sites to gene promoters:

$ bedmap --echo --echo-map --delim '\t' promoters.bed tfbs.bed > answer.bed

Each line of the file answer.bed contains a promoter region, its associated gene ID, and any TF binding sites that overlap the gene's promoter.

[david.bolumar]

OK, I am going for that, thanks a lot for the help!

Answer 2

[david.bolumar]

Hi!

I actually have the regions on the genome where certain transcription factors bind, thanks!