[Biopython-dev] Error in SeqFeature.CompoundLocation parsing NCBI efetch format

[Brynjar Smári Bjarnason]

On 6 December 2013 09:03, Brynjar Smári Bjarnason <binni at binnisb.com> wrote:

> On 5 December 2013 19:08, Brynjar Smári Bjarnason <binni at binnisb.com>wrote:
>
>> Thanks, will look at this when I'm at the computer :-)
>> On 5 Dec 2013 19:06, "Brynjar Smári Bjarnason" <binni at binnisb.com> wrote:
>>
>>> I'll ask one who knows but I think I could skip using the bonds. Can you
>>> suggest how I can ignore the bonds in efetch response, or the parser?
>>>
>>> Thanks a lot for looking at this!
>>> On 5 Dec 2013 18:12, "Peter Cock" <p.j.a.cock at googlemail.com> wrote:
>>>
>>>> On Thu, Dec 5, 2013 at 4:46 PM, Peter Cock <p.j.a.cock at googlemail.com>
>>>> wrote:
>>>> >
>>>> > Not to worry - the site did respond when I retried a bit later, and
>>>> > I can reproduce the parser error:
>>>> >
>>>> >>>> from Bio import SeqIO
>>>> >>>> r = SeqIO.read("1MRR_A.gp", "genbank")
>>>> > BiopythonParserWarning: Couldn't parse feature location:
>>>> > 'join(bond(84),bond(115),bond(118),bond(238))'
>>>> > BiopythonParserWarning: Couldn't parse feature location:
>>>> > 'join(bond(115),bond(204),bond(238),bond(241))'
>>>> > BiopythonParserWarning: Couldn't parse feature location:
>>>> > 'join(bond(194),bond(272))'
>>>> > ...
>>>> > ValueError: CompoundLocation should have at least 2 parts
>>>>
>>>> The problem is the bond locations, and in particular while the
>>>> parser gave up on the ones with a warning, it fell over the
>>>> single bond entry, bond(196).
>>>>
>>>> This is partly due to a change in the use of the bond term,
>>>> which used to be a compound entry like bond(194,272).
>>>> Also the GenBank parser was and is primarily used on
>>>> nucleotide sequences rather than GenPept files which are
>>>> occasionally more weird (like here!).
>>>>
>>>> A short term hack would be to strip out the bond term
>>>> (with a warning) and parse the remainder as a simple
>>>> join or single residue accordingly.
>>>>
>>>> Would that work for you - do you need the bond bit?
>>>>
>>>> Peter
>>>>
>>>
> I believe for our part that leaving the bond bit out is fine so your patch
> should work well.
>
> Any suggestions on a good way to apply this patch? Should I build
> Biopython from that branch or clone latest stable and apply the patch
> before building?
>
> Thank you
>
> Brynjar
>

The patch works for me! Thank you.

I cloned the official Biopython and applied your commit as patch before
building. My gis' that were failing now work.

Brynjar