Extract PMIDs from a gene or protein ID

Question

[shelly.deforte]

Given a uniprot ID, I am trying to automatically extract related pubmed IDs (PMIDs) from pubmed. I can map the UniProt ID to something NCBI can understand. For instance, UniProt ID O14733 can be mapped to GI:6831583 and then you can launch a search from http://www.ncbi.nlm.nih.gov/protein/O14733 to see the associated pubmed articles with the URL http://www.ncbi.nlm.nih.gov/pubmed?linkname=protein_pubmed_weighted&from_uid=6831583.

I have never used ncbi's e-utils, so it may be a very simple modification to be able to fetch these articles automatically, but I can't figure it out. My best guess was http://eutils.ncbi.nlm.nih.gov/entrez/eutils/esearch.fcgi?db=pubmed&LinkName=protein_pubmed_weighted&from_uid=6831583, but this returns nothing.

Basically, given an ID such as 683583, I want to return a list of PMIDs. I would rather do this in python if possible. Any suggestions?

[Ashutosh Pandey]

Similar posts:

• edirect: Entrez Unix Command line
• Retrieve Pubmed Ids

Answer 1

[David W]

If you use the link EUtil with dbfrom="protein", db="pubmed" you'll get a list of pmids associated with that protein.

You can then use esummary or efetch on those pmids.

[shelly.deforte]

Thanks, that's what I needed!

Answer 2

[shelly.deforte]

For completeness, here's what I worked out with David W's help. The construction of the URL is such:

http://eutils.ncbi.nlm.nih.gov/entrez/eutils/elink.fcgi?dbfrom=protein&db=pubmed&id=215274019&linkname=protein_pubmed_weighted

This is how I was able to retrieve the records in Biopython:

protein_ID = "215274019"

handle = Entrez.elink(db="pubmed", dbfrom="protein", id=protein_id, linkname="protein_pubmed_weighted")
record = Entrez.read(handle)

for PMID in record[0]['LinkSetDb'][0]['Link']:
    print PMID['Id']

[nils.rudqvist]

Hey, I am trying to do the same thing basically. I have a q. D o you know the differences between the different linknames, i.e., protein_pubmed_weighted, protein_pubmed, and protein_pubmed_refseq?

Best,
Nils

Answer 3

[asking.for.help]

Depending on the scope of your project, you might want to directly download NCBI's look-up table between EntrezIDs and PubmedIDs and integrate this table into your workflow.

ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2pubmed.gz

I like this table because it enables easy (and computationally fast) filtering; e.g.: exclude papers, which cover 100s -1000s of different genes (and usually thus do not reveal gene-specific biology). e.g.: find genes, which are only mentioned together with your genes of interest

[shelly.deforte]

This looks like a great resource, though I don't think I can easily map my UniProt IDs to genes, and I think it might change the coverage of the papers if I did. Still, I'm definitely going to bookmark this folder, thanks!