Hg38 TSS .bed/.gtf file for chipseq analysis

Question

[satishedu]

Hi Everyone,

How can I obtain a bed file with the TSS region? I am performing a chipseq analysis and to see the read covrage around the TSS using deeptools , I would like to have the hg38 TSS region .bed/gtf file. I googled a lot for this and found good solution for the hg19 assembly. It was hard for me to find the right solution for hg38 assembly. I kindly request someone to guide me to get the required files.

Thank you.

Satish

Answer 1

[Devon Ryan]

We normally download GTF files from Ensembl and then convert using UCSC tools:

awk '{if ($$3 != "gene") print $0;}' file.gtf \
| grep -v "^#" \
| gtfToGenePred /dev/stdin /dev/stdout \
| genePredToBed stdin genes.bed

It's not pretty, but it works.

I should note that the next release of deepTools will accept GTF files, which should make everyone's life easier :)

[satishedu]

Thanks Devon, I will try this approach. I would also like to ask what kind of .bed file should I be exactly having to know the read coverage intensity of my chip seq around TSS. I am working chipseq on FOXO transcription factor.

[Devon Ryan]

You want at minimum a 6 column BED file, with entries for each transcript. There are easier ways of making that than what I showed to you, but if you wanted a BED12 file for future use (e.g., the next version of deepTools) then you'd already be good to go with it.

[Jeffin Rockey]

Devon, can you please provide two more information

Why two dollar signs used in $$3 != "gene" instead of just $3. I am quite curious to know whether $$ have any different meaning.

Also,if I am to get a bed file with TSS regions alone , could you please suggest any tweaks required with the given script. (I tried the given script, the output bed seemed to have lot more information than just TSS co-ordinates).

[Devon Ryan]

This was originally being run inside of a Makefile, so it needed to be escaped properly. That wouldn't be needed otherwise. I probably should have noted that when I posted this originally.

[Jeffin Rockey]

Devon, Thanks a lot for clarifying.

Regarding the other query, is there a robust way that you would recommend to get a bed/gtf file for TSS alone?

[Devon Ryan]

If you have BED file then it's a bit easier. It'd be something like:

awk '{if($6 == "-") {$2 = $3 - 1} else {$3 = $2 + 1} print}' input.bed > output.bed

For a GTF file it'd be something like:

awk '{if($3 == "transcript") {if($7 == "-") {$2 = $3} else {$3 = $2} print}}' input.gtf > output.gtf'

If there are comment lines then they'd need to be removed, either with grep or directly in awk.

[Jeffin Rockey]

Thanks Devon. The two one liners provided would be of great help to me indeed.

Jf