pythonic way to get gene_id, gene_symbol from given list of gene, transcript co-ordinates

Question

[badribio]

Any quick methods to get ensemble gene_id, gene_symbol for a list of co-ordinates (transcript), this post A: Identify gene symbols given a list of chromosome positions talks about ucsc only.

[finswimmer]

Hello,

you could use ensembl's REST-Api for this, e.g. the Overlap endpoint.

fin swimmer

[badribio]

any snippet to query rest api using python?

Answer 1

[Devon Ryan]

I'd be surprised if this couldn't be done with biomart.

In fact, here is a very simple example using GRCh38.

[badribio]

biomart has a limit of 500 queries, (I may be wrong here) I have a lot of lines as this is transcript level co-ordinates

[Devon Ryan]

BTW, if for some reason you really want a python-based solution then download a GTF file and:

pip install deeptools

then in python

from deeptoolsintervals import GTF
anno = GTF("foo.gtf", transcriptID="gene_id", transcript_id_designator="gene")
anno.findOverlaps("chr1", 1, 1000)

That will get you the gene_id field and coordinate information. The python wrapper doesn't allow access to the symbol, so you'd need to just download the mapping from biomart.

If you don't want to perform a bunch of remote queries then something along those lines would work. I never really intended for others to use that python module, but if you ever want to it's documented here.

[badribio]

Thanks I will try this out, I need python solution as I need to modify a pipeline which has been written using python. else bedtools was my first choice, having said that pybedtools should also do the job I am not wrong.

[Devon Ryan]

Yeah, pybedtools would have been my other suggestion.

[Devon Ryan]

At that point don't you have the transcript IDs? Then you don't need to look anything up with coordinates, you just need to convert the transcript to gene ID (also available on biomart).

[badribio]

Nope, I have the co ordinates just like output from tophat junctions.bed file.

[Devon Ryan]

I expect that bedtools intersect and a bit of awk will turn out to be the simplest solution :P