[Biopython-dev] WIT and KEGG

[Tarjei S Mikkelsen]

> > >  Instead I suggest that you either run the input through
> > > File.SGMLHandle or File.SGMLStripper before you pass the
> > > WIT record to KEGG.Enzyme.Parser OR write a separate Parser
> > > class in your WIT module that wraps a
> ParserSupport.SGMLStrippingConsumer
> > > around KEGG.Enzyme._Consumer.
> > >
> >   The problem is I'm experimenting with a filter to strip out junk ( not
> > necessarily html ) between records.
> > The motivation is that I've had Martel fail on just an extraneous line
> feed.
> > Somehow the idea of chaining two filters together trips a watch for bugs
> > alarm in my mind.

Sure, for experimentation that's fine, but I'd prefer to keep it the way it
is in the distribution version. Especially because the HTML versions of
these records are full of other markup _in_ the record that has to be
cleaned out anyway - and adding regexps for all of those would be a mess.

> > > >   The format failed halfway through the file.  I think the
> problem is
> > the
> > > > order of entries.  The format specifies GENES before MOTIF but
> > > > this order is
> > > > reversed in the test file.  Maybe the format should be less
> sensitive
> to
> > > > order ,where it doesn't convey information.
> > >
> > >  Yeah, the entries are supposed to come in a specified order, but even
> > > the KEGG people don't follow that rule. I've committed a change to
> > > KEGG.Enzyme.enzyme_format.py that assumes very little about entry
> > > ordering. If that's the error, it should work for you now.
> > >
> >
> > Now its stopping on files with db links like this example:
> >
> >             PIR: B49338  B49935  E64239  KIECAA
> >
> > These are quibbles but the computer doesn't understand quibbles:).

 Yeah, I missed this case because it doesn't appear in KEGG. I've committed
another change which appears to deal well with it.

 Btw, I'm going away for a couple of weeks, so I'll won't be very responsive
during that time. But I'm planning to bring my laptop to do some more
experiments with reaction/pathway classes.

 take care,

 Tarjei