[Biopython-dev] NCBI DTD File

[Peter Cock]

Hi Lev,

Which version of Biopython do you have, and which GI number(s) fail?

The very fact the problem tag was "Error" suggests it was actually
an error message, not a sequence record - perhaps a temporary error?

This worked for me:

from Bio import Entrez
Entrez.email = "..."
handle = Entrez.efetch(db="protein", id="12345678", retmode="xml")
record = Entrez.read(handle, validate=True)
handle.close()
print(record)

Using some id values like "1" could give an "empty" XML record,
which to me looks like an NCBI bug:

<?xml version="1.0"?>
 <!DOCTYPE GBSet PUBLIC "-//NCBI//NCBI GBSeq/EN"
"http://www.ncbi.nlm.nih.gov/dtd/NCBI_GBSeq.dtd">
 <GBSet>

</GBSet>

This is parsed as [] which is reasonable (empty list).

Other values like "0" and "-1" give an HTTP Error 400: Bad Request
(which is good - a nice clear and explicit error).

See also:

Peter


On Fri, Sep 4, 2015 at 8:16 PM, Lev Tsypin <ltsypin at uchicago.edu> wrote:
> Hi Peter,
>
> This is me trying to get protein sequences from the protein database. I have
> a gi code in the variable 'gi' that I pass into the Entrez.efetch function.
> Specifically, I use:
>
>         handle = Entrez.efetch(db='protein', id=gi, retmode='xml')
>         record = Entrez.read(handle)
>
> Best,
> Lev
>
> On Fri, Sep 4, 2015 at 11:12 AM, Peter Cock <p.j.a.cock at googlemail.com>
> wrote:
>>
>> Hi Lev,
>>
>> Which database was this with? Each has somewhat different XML behaviour./
>>
>> The NCBI have been quite good about versioning the DTD files -
>> normally they add new files rather than edit an existing DTD file. So
>> unless you've had a warning from Biopython there should be no reason
>> to download a new DTD file.
>>
>> Peter
>>
>> On Fri, Sep 4, 2015 at 3:44 PM, Lev Tsypin <ltsypin at uchicago.edu> wrote:
>> > Hi all,
>> >
>> > I am encountering this error when using Bio.Entrez:
>> >
>> > Bio.Entrez.Parser.ValidationError: Failed to find tag 'Error' in the
>> > DTD. To
>> > skip all tags that are not represented in the DTD, please call
>> > Bio.Entrez.read or Bio.Entrez.parse with validate=False.
>> >
>> > I've found a discussion of the same issue from about a year ago, so I
>> > figure
>> > the the NCBI updated their DTD file in a strange way. I found several
>> > solutions: would you recommend that I download the new DTD file into my
>> > local copy of Biopython or run Entrez.read with validate=False?
>> >
>> > Best regards,
>> > Lev Tsypin
>> >
>> > _______________________________________________
>> > Biopython-dev mailing list
>> > Biopython-dev at mailman.open-bio.org
>> > http://mailman.open-bio.org/mailman/listinfo/biopython-dev
>
>