[BioRuby] Can I get more information on BioRuby

[Rob Syme]

Hi Dave

Don't worry, these are appropriate questions for someone new to bioruby.
Good on you for digging through the source code.

The NCBIDB class is located at
lib/bio/db.rb<http://github.com/bioruby/bioruby/blob/master/lib/bio/db.rb#L233>
.
Be sure to have a look at the top of the file for a great overview of how
the database classes are constructed, written by ktym<http://github.com/ktym>
.

To track down the method of autodetection, have a close look at the
lib/bio/io/flatfile.rb<https://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile.rb#L80-111>
 and lib/bio/io/flatfile/autodetection.rb<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile/autodetection.rb>
.


*Step 1) Check to see if a dbclass has been provided*
If no database class is provided to Bio::FlatFile.open, it falls
through to _open_file(dbclass,
filename, *arg)<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile.rb#L155-165>,
where dbclass is nil, which then creates a new Bio::FlatFile
instance<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile.rb#L158>
with
the first argument (dbclass) as nil.

If you look at the FlatFile initialize method, you'll see that the autodetect
method is called<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile.rb#L233-237>if
dbclass is nil.
Inside the autodetect method, a new Autodetect object is
created<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile.rb#L429>
which
does the work you are interested in. Note that the class method "default"
is called, which returns an Autodetect object, but does a lot of setting up
first<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile/autodetection.rb#L348>
.


*Step 2) Set up the default FlatFile::Autodetect object*
The setup is mainly concerned with
adding<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile/autodetection.rb#L363>
a
large number of RuleTemplate instances to the @elements array. Each
detectable format type has it's own RuleTemplate.
The RuleTemplates provide three key pieces of information:
 - a name<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile/autodetection.rb#L46>
 - a list of rules that are less
important<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile/autodetection.rb#L50-55>(checked
later) that the current rule
 - a guess method<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile/autodetection.rb#L76-78>,
which, when given a string, returns a dbclass if the string looks like it
belongs to that particular rule.

There are different types of RuleTemplates. The simplest is a
RegRegexp<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile/autodetection.rb#L143-173>,
which has a very simple guess method, which simply matches the given string
given to a regular
expression<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile/autodetection.rb#L170-173>
.

The RuleTemplates also need a list of rules that are less important so that
the final array of rules can be topologically
sorted<http://www.ruby-doc.org/stdlib-1.9.3/libdoc/tsort/rdoc/TSort.html>,
returning a list rules with the most important at the top.

There is a large array created in Autodetect.make_default which contains a
bunch of rules that are added to the @elements array (to be tsorted later).

*Step 3) Go through the first 31 lines, and match against each of the
RuleTemplates*
Way back in the FlatFile#autodetect method, a dbclass is pulled out of the
FlatFile::Autodetect#autodetect_flatfile
method<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile.rb#L430>
.
The FlatFile's BufferedInputSteam that was created
earlier<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile.rb#L161>
is
pulled out so that the Autodetect object can peek inside.
It then iterates over the first 31 lines of the
file<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile/autodetection.rb#L333-342>,
taking each line and trying it out against each
rule<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile/autodetection.rb#L309>.
If the guess returns non-nil (it will be a dbclass), that dbclass bubbles
all the way back up to the autodetect method in
FlatFile<http://github.com/bioruby/bioruby/blob/master/lib/bio/io/flatfile.rb#L430>
.


I hope that's right - I've not contributed any code here. I might be way
off.
It does raise the question "How do I get a new dbclass to be recognised by
FlatFile.open?" I'm not really sure, but I'm confident there will be
someone here who does know. Any takers?

-r
*
*



On Fri, Mar 8, 2013 at 3:30 AM, dave thorneycroft
<dthorneycroft at gmail.com>wrote:

> Hello,
>
> I am a novice Ruby programmer, ex-biologist and experienced developer in
> other languages (sorry Adobe Flex), but I have seen the light and  I want
> to contribute to BioRuby.  I am looking through the BioRuby source code and
> find somethings confusing (of course I'm a novice !).  Could anyone point
> me in the direction of any tutorials which wil help me get a better
> understanding of the code,  I've seen the stuff on BioRuby.org.  Anything
> that would explain  but how its structured and coded not how to use it to
> solve a problem.
>
> A couple of things are really puzzling me right now.
>
> Number one, please don't laugh , where is the NCBIDB class located in the
> source (I see the GenBank class inherits NCBIDB ; but I cannot find the
> class source ?). NCBIDB is a class right ?
> Number two, could anyone explain how the Bio::FlatFile 'automagically'
> recognizes each database class?
>
> Any pointers would be great.  Many thanks for your time, I really
> appreciate any comments.
>
> Regards
> Dave
> _______________________________________________
> BioRuby Project - http://www.bioruby.org/
> BioRuby mailing list
> BioRuby at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/bioruby
>