[Biopython] Access Entrez gene DB using rettype 'gb'

[David Jacobs]

>
> > Thanks for the info. There seems to be a huge disconnect between what I
> want
> > to do and what this library is letting me do. It seems like there should
> be
> > a really simple way to look up bacterial gene sequences by their names,
> and
> > it's disappointing that that's not the case.
> >
> > Every workaround I've tried has also failed.
> >
> > For example, I've downloaded the full CT18 genome from the FTP server and
> > parsed it using SeqIO. The problem is that SeqRecord doesn't give me an
> > accessor to the "name" attribute of the sequence, as it would appear in
> the
> > gene database.
>
> You'll have to give me more to go on - what did you download by FTP,
> a FASTA file, GenBank? How about giving the URL and an example
> of the "name" you want to use.
>

In this case, I downloaded the file Brad listed earlier:

ftp://ftp.ncbi.nih.gov/genomes/Bacteria/Salmonella_enterica_serovar_Typhi_CT18_uid57793/NC_003198.ffn

The "name" I'd like back is the name listed as "symbol" at
http://www.ncbi.nlm.nih.gov/gene when I query "ct18 fliC"--in other words, I
want to search the genome file I have for "fliC", and not just the
human-readable description. It seems to me like the "name" attribute for
SeqRecord would be a useful place to put this, especially since right now,
"name" is just a duplicate of the information in "id".

This already works for protein entries. See:

https://github.com/biopython/biopython/blob/master/Bio/SeqRecord.py#L322

The thing is, the human-readable description of each gene is already
annotated in the genome FASTA file I downloaded. I just need the symbol, as
it's easily searchable and more canonical.

> What's more, if I search the gene database for a name, I do,
> > in fact, get an ID back. But that ID has no information about the start
> and
> > stop indices for my sequence, so I can't use that information in
> conjunction
> > with my downloaded genome.
>
> Have you looked at EInfo? It is for cross referencing between the different
> Entrez databases.
>

I'll have a look.


> > Further still, if I try to query the gene
> > database for my gene's full information (using the ID that I grabbed from
> > esearch(db=gene ...)), I get back data formatted in a way that BioPython
> > can't parse.
>
> Are you talking about using EFetch here? Which database? The valid
> combinations of retmode and rettype change according to this. See e.g.:
> http://www.ncbi.nlm.nih.gov/corehtml/query/static/efetchseq_help.html


I'm using ESearch. As I say, I'm trying to query the "gene" database using a
name ("ct18 fliC"). I do get back just one entry, and it gives me the
correct GID. When I try to query "gene" using this ID--in order to get the
start and stop indices--the best I get is:

http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=gene&id=1248507&retmode=txt

<http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=gene&id=1248507&retmode=txt>Which
I don't know how to parse. (Again, I just want the start and stop
positions.)

 > This is a touch aggravating.

> >
> > What am I missing?
> >
>
> The NCBI Entrez documentation is definitely sparse :(
>
> If all you want to do is get the nucleotide sequence for bacterial
> genes then I do suspect working with the FASTA or GenBank files
> would be easier than using Entrez (as Brad suggested earlier).
>

I'd rather not do this manually, though. It seems like BioPython should make
tedious tasks like this easy and sustainable.


> Can you give a specific example - couple of gene names you want,
> and desired answer (the sequence want to find for them)? Sean did
> ask earlier - this really would and we'd be better able to help you.


For the example I've given over the last couple of e-mails, this is the
sequence I want:

http://www.ncbi.nlm.nih.gov/nuccore/NC_003198?report=fasta&from=2011173&to=2012693&strand=true

It's directly linked to as from the fliC page in the "gene" database. (From
the link labeled "FASTA".)

Does that make things clearer?

David