How to map sub-cellular localisation to enteries in uniprot database fasta file.

Question

[wl284]

I have a dataset of proteins that I have blasted against the uniprot-swissprot database.

I'd now like to identify which proteins are likely to have a mitochondrial sub-cellular localisation based on the sub-cellular localisation of their best blast hit in the swiss-prot database.

The fasta headers of the uniprot proteins look like this:

">sp|Q64602|AADAT_RAT Kynurenine/alpha-aminoadipate aminotransferase, mitochondrial OS=Rattus norvegicus GN=Aadat PE=1 SV=1"

I have found a gene ontology mapping file (link below) but the fasta headers don't contain the GO IDs necessary to map them. ftp://ftp.ebi.ac.uk/pub/databases/GO/goa/external2go/uniprotkb_sl2go

Is there some intermediate file that I need to use and does anyone know where to find it? Any help would be appreciated.

Answer 1

[Pierre Lindenbaum]

using xslt:

$ awk -F '|' '/^>/ {printf("%s\n",$2);}' input.fa | while read ACN ; do curl -s "https://www.uniprot.org/uniprot/${ACN}.xml"| xsltproc transform.xsl - ; done

Q64602  Mitochondrion

with transform.xsl:

[wl284]

Thanks, that's awesome.

[Michael]

I changed the protocol to HTTPS, otherwise the response could be empty, because Uniprot move to https and sends a document moved header.

[Pierre Lindenbaum]

yes I saw it this morning ! :-D https://github.com/lindenb/jvarkit/commit/5f6b66bc05201d2d543e1b1214640dd5c84051f8