[Bioperl-l] Bio::Tools::Glimmer

[Mark Johnson]

On 9/19/06, Torsten Seemann <torsten.seemann at infotech.monash.edu.au> wrote:

> Glimmer3 produces two separate files: XXX.detail and XXX.predict.
> The Glimmer3 .detail file is similar to the first part of the Glimmer
> 2.x first part. The Glimmer3 .predict file conveys the same information
> as the second part of a Glimmer2 file, but in a totally different format!

Also, Glimmer2 only analyzes the first sequence in the input fasta
file, whereas Glimmer3 will happily take many sequences, and even tell
you which predictions go with what sequence in the output, with
markers that looks like fasta headers: > sequence_name ...

Apart from the technical issues of parsing the different formats,
there is an up-front design question?  Should both the open reading
frames and final gene predictions be processed into features, or just
the final predictions?  If the former, that is going to make a single
abstract interface tricky, as Glimmer3 has two input files that need
to be parsed, where the rest would have one.  Though, if only the
predictions get parsed, well, that solves that problem.  8)

...
...
...

> It doesn't currently work with my GlimmerHMM output, as the module
> expects a version number, which my output does not have - but I will fix
> that in CVS today.

Sadly, Glimmer3 seems to output no version info to either the .predict
or .detail file.