Entrez API returns results different from main search?

Question

[Philipp Bayer]

I'm having a relatively baffling problem - I'm trying to access the info for a bunch of BioSample IDs via the Entrez API in Python using Biopython v1.68, here's a minimal example. I'm looking for BioSample SAMEA2467098, URL: https://www.ncbi.nlm.nih.gov/biosample/?term=SAMEA2467098

This should be Brassica napus, cv. Beluga

In Python:

from Bio import Entrez
Entrez.email = 'SNIP'
sample = 'SAMEA2467098'
handle = Entrez.efetch('BioSample', id=sample, retmode='text')
print(handle.readlines())
['1: Non-tumor DNA sample from blood of a human participant in the dbGaP study "An APOBEC Cytidine Deaminase Mutagenesis in Human Cancers"\n', 'Identifiers: BioSample: SAMN02467098\n', 'Organism: Homo sapiens\n', 'Attributes:\n', '    /submitter handle="NHGRI_APOBEC_CytidineDeaminase"\n', '    /biospecimen repository="NHGRI_APOBEC_CytidineDeaminase"\n', '    /study name="An APOBEC Cytidine Deaminase Mutagenesis in Human Cancers"\n', '    /study design="Cross-Sectional"\n', '    /biospecimen repository sample id="TCGA-BP-5199-11A-01D-1429-08"\n', '    /submitted sample id="TCGA-BP-5199-11A-01D-1429-08"\n', '    /submitted subject id="TCGA-BP-5199-11A-01D-1429-08"\n', '    /study disease="Neoplasms"\n', '    /tissue="blood"\n', '    /analyte type="DNA"\n', '    /is tumor="No"\n', '    /subject is affected="Yes"\n', '    /molecular data type="SNV (.MAF)"\n', 'Accession: SAMN02467098\tID: 2467098\n', '\n', '\n']

Suddenly it's human, and the accession changed from SAME A 2467098 to SAM N0 2467098, which is indeed human: https://www.ncbi.nlm.nih.gov/biosample/?term=SAMN02467098

What am I doing wrong here? It works for most of my other BioSample IDs!

Edit: It's not a Biopython problem, using the API manually gives the same result: https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=biosample&id=SAMEA2467098&retmode=text

[GenoMax]

May be worth creating a ticket at NCBI support. Looks like some sort of a bug in their code.

[Philipp Bayer]

I dropped them a message late last night, looks like they'll file it as a bug, will update here

[Zorino]

Any update on this issue ?

The bug doesn't seem to be fixed yet.

[Philipp Bayer]

I haven't heard back anything, but just going over the middle step of using the returned ID works

Answer 1

[Philipp Bayer]

Hm, I think this has to do with the way Entrez handles IDs, if I use esearch first to get the BioSample's ID number it works:

This works, using esearch first:

handle = Entrez.esearch('BioSample', term='SAMEA2467098', id='text')
record = Entrez.read(handle)
record
{u'Count': '1', u'RetMax': '1', u'IdList': ['3769633'], u'TranslationStack': [{u'Count': '1', u'Field': 'All Fields', u'Term': 'SAMEA2467098[All Fields]', u'Explode': 'N'}, 'GROUP'], u'TranslationSet': [], u'RetStart': '0', u'QueryTranslation': 'SAMEA2467098[All Fields]'}

Using the new ID now in efetch:

handle = Entrez.efetch('BioSample', id=3769633, retmode='text')
handle.readlines()
['1: Beluga_PBY013; Beluga\n', 'Identifiers: BioSample: SAMEA2467098; SRA: ERS436855\n', 'Organism: Brassica napus\n', 'Attributes:\n', '    /sample name="ERS436855"\n', 'Accession: SAMEA2467098\tID: 3769633\n', '\n', '\n']

We're back at Beluga! But I'm still not 100% sure why this works the way it does, possibly the ID I'm entering clashes with the human entry or gets shortened for search.

[Sej Modha]

esearch UNIX utils seems to work

esearch -db biosample -query "SAMEA2467098"|esummary

[Philipp Bayer]

Interesting, I can confirm this! I reckon because this one asks or a 'query' while I'm supplying an 'ID' above?

[Sej Modha]

Yes, I think you're right because if I supply ID then efetch returns the human results.

efetch -db biosample -id "SAMEA2467098"

1: Non-tumor DNA sample from blood of a human participant in the dbGaP study "An APOBEC Cytidine Deaminase Mutagenesis in Human Cancers" Identifiers: BioSample: SAMN02467098 Organism: Homo sapiens

[Peter]

I'd guess that NCBI Entrez Fetch is being 'helpful' by accepting the the sample name, when it really wants the numerical identifier - and this is going wrong. I'd email the EUtilities help team about this if I were you.

Answer 2

[DCGenomics]

Two things,

NCBI Education Group has been offering webinars on this lately.

Check out some of the recent ones at:

https://www.ncbi.nlm.nih.gov/home/coursesandwebinars.shtml

Also, if you put in an issue here, we'll try to help you with a solution:

https://github.com/NCBI-Hackathons/EDirect_EUtils_API_Cookbook