[Buffalo-pm] Statistics With Perl...

Shankar, Ganesh Ganesh.Shankar at RoswellPark.org
Thu Jan 19 07:46:00 PST 2006


 Dan,

 Yes, that's pretty much my opinion. The work under the hood will be done by existing code, be it Perl or not.  The reason I mentioned that bioinformatists will move away from Perl is because there is no pluggable framework.  With your initial research (good job,by the way), I can see that the major pieces are there, but they need to be glued together. I haven't looked into your links at all, but will do so soon as I get back from vacation.

 I think that there are going to be two classes of users - biologists and bioinformaticians.  The bioinformatics people will be satisfied if we just give them a consistent framework and api with command line tools.  The biologists will be at the other extreme where they won't care what language,module is doing the work, they just care that the tools support their workflow.  For them, the GUI is paramount, next to speed.  The major piece that's missing is the GUI.  The tools also have to interact with public databases (www.ncbi.nlm.nih.gov and http://www.ensembl.org/) but that's a forte of Perl and should be manageable.

 So, I see 3 challanges: build/integrate the framework, GUI, and lastly port existing statistical algorithms to the framework.  If we can do this in Perl, then it should be faster than what I've experienced so far.  Maybe we can propose this as a business idea for the next School of Management's yearly contest.  This is precisely the kind of approach I was advocating as an alternative to the multi-tier, grid enabled, bioinformatics environment that is being developed.

Definetely worth thinking about.

-Ganesh



-----Original Message-----
From: buffalo-pm-bounces at pm.org on behalf of DANIEL MAGNUSZEWSKI
Sent: Wed 1/18/2006 11:17 PM
To: buffalo-pm at mail.pm.org
Subject: [Buffalo-pm] Statistics With Perl...
 
Ganesh,

Correct me if I'm wrong, but you mentioned last night that
Bioinformatics is beginning to focus more on the statistics and
probability of test results being correct and less on the actual lower
level programming aspects (such as using Perl to do sequencing and give
you the test results, etc). This is because of the large code base that
is already in place. You then mentioned that because of this, people
within Bioinformatics (in general) will begin to move away from Perl,
and begin to use a language like R (http://www.r-project.org/). After we
mentioned how there was probably a wrapper module for R, I poked around
CPAN and found a module, Statistics::R, that will allow you to interact
(create, start, stop, etc) with an R interpreter through Perl.

Statistics::R - Controls the R (R-project) interpreter through Perl.

http://search.cpan.org/~gmpassos/Statistics-R-0.02/lib/Statistics/R.pm

Below is another R module I found, but it states that it doesn't
implement all functions of R - so I'm assuming that it can be quite
limiting:

PDL::R::math - routines from statistics language, R, for PDL

http://search.cpan.org/~cavanaugh/PDL-R-math-0.12/rmath.pd

Also, I'm not sure if you use programs like MATLAB, but there is an open
source version of it called Octave (http://www.octave.org). As with most
everything, there are some Perl modules available to help you interact
with it:

Inline::Octave - Inline octave code into your Perl

http://search.cpan.org/~aadler/Inline-Octave-0.22/Octave.pm

I have even found a package available that will provide an interface
between R and Octave - properly named, ROctave (
http://www.omegahat.org/ROctave/ ). The same group that releases
ROctave, has a package called RSPerl ( http://www.omegahat.org/RSPerl/ )
that "provides a bidirectional interface for calling R from Perl and
Perl from R". Essentially, these tools provide statistical functionality
to Perl that is not natively present.

Does any of this seem useful for what Biologist need to do? Would this
integrated "framework" help for creating better/more effective tools? 

Hopefully some of this helps...

-Dan


_______________________________________________
Buffalo-pm mailing list
Buffalo-pm at pm.org
http://mail.pm.org/mailman/listinfo/buffalo-pm



This email message may contain legally privileged and/or confidential information.  If you are not the intended recipient(s), or the employee or agent responsible for the delivery of this message to the intended recipient(s), you are hereby notified that any disclosure, copying, distribution, or use of this email message is prohibited.  If you have received this message in error, please notify the sender immediately by e-mail and delete this email message from your computer. Thank you.


More information about the Buffalo-pm mailing list