headers
Catherine Letondal | 4 Jan 2005 17:23
Picon
Picon
Favicon

Re: Numeric + Lapack/Blas/ATLAS [Re: [BioPython] Installation help]


On Dec 20, 2004, at 5:30 PM, Catherine Letondal wrote:

>
> On Apr 7, 2004, at 12:39 AM, Brad Chapman wrote:
>
>> Hi Jim;
>> [Condensing both answers into a single mail]
>>
>>> I'm pulling together the dependencies for biopython
>>> and need to know whether I should download "numeric"
>>> or "numarray".
>>
>> Numeric. Although Numarray is supposed to be API compatible to some
>> extent with Numeric, I don't believe that anyone has fully tested
>> that out. Numeric should work for now -- at some point in the future
>> someone will likely go through and ensure everything works and we'll
>> move on to Numarray.
>>
>
> Hi,
>
> Has someone succesfully installed Numeric on MacOSX/PPC/G4? I have 
> some problems while installing this package, because it needs another 
> package (Lapack):
>
> % python setup.py install
> [...]
> building 'lapack_lite' extension
> gcc -fno-strict-aliasing -Wno-long-double -no-cpp-precomp
(Continue reading)

Permalink | Reply | 4 comments |
headers
K. Arun | 4 Jan 2005 21:32
Picon

Re: [BioPython] PDBparser and NMR pdb file

On Thu, 16 Dec 2004 18:41:03 +0100, Edoardo Saccenti
<saccenti <at> cerm.unifi.it> wrote:

> does anyboy know if is possible to handle PDB files
> of structures obteined via NMR spectrometry?
> E.G. obtein models of various conformers in a file?

I wrote some quick and dirty code to do this a while ago - I'm sure
the Bio.PDB module code can be used more effectively than it is below.

-arun

## --
#!/usr/bin/env python

"""
extract-model.py :
Script to spit out arbitrary models from a PDB NMR structure.
Note that only the MODEL, ATOM, TER, and ENDMDL lines are
retained. All header detail is stripped. The lines are printed to
stdout, and will need to be redirected to an appropriate file.

Usage:
extract-model.py PDB-ID PDB-filename model-no-to-extract

K. Arun <karbak <at> cmu.edu>

"""

import sys, string, re, os
(Continue reading)

Permalink | Reply | 0 comments |
headers
Ivan Rossi | 5 Jan 2005 11:43
Favicon

[BioPython] numeric vs numarray

>> On Apr 7, 2004, at 12:39 AM, Brad Chapman wrote:
>>> 
>>> Numeric. Although Numarray is supposed to be API compatible to some
>>> extent with Numeric, I don't believe that anyone has fully tested
>>> that out. Numeric should work for now -- at some point in the future
>>> someone will likely go through and ensure everything works and we'll
>>> move on to Numarray.

People I trust did some comparison between numeric and numarray and told me 
that, despite its claim Numarray is much much slower than numeric even on 
large matrices (10kx10k), even by an order of magnitude. So they plan to stay 
with numeric for a long time to come.

Any other evidence in favor or against numarray from people in the list?

--
  Ivan Rossi, Ph.D. - ivan <at> biodec.com, ivan.rossi3 <at> unibo.it
  BioDec s.r.l., Via Fanin 48, I-40127 Bologna (Italy)
  Phone: +39-051-4200321 - fax: +39-051-4200317 - web: www.biodec.com
_______________________________________________
BioPython mailing list  -  BioPython <at> biopython.org
http://biopython.org/mailman/listinfo/biopython
Permalink | Reply | 2 comments |
headers
Simon Burton | 6 Jan 2005 00:25

Re: [BioPython] numeric vs numarray

Ivan Rossi wrote:

>
> People I trust did some comparison between numeric and numarray and 
> told me that, despite its claim Numarray is much much slower than 
> numeric even on large matrices (10kx10k), even by an order of 
> magnitude. So they plan to stay with numeric for a long time to come.
>
> Any other evidence in favor or against numarray from people in the list?
>

There is a discussion about this issue (with some benchmarks) right now 
on numpy-discussion:
http://sourceforge.net/mailarchive/forum.php?thread_id=6274118&forum_id=4890

My current employer has decided to go with numarray for a new 
statistical machine learning project. We hope to contribute to work in 
improving numarray performance, as needed.

bye,

Simon.

--

-- 
Simon Burton, B.Sc.
Licensed PO Box 8066
ANU Canberra 2601
Australia
Ph. 61 02 6249 6940
http://arrowtheory.com
(Continue reading)

Permalink | Reply | 0 comments |
headers
Michiel Jan Laurens de Hoon | 6 Jan 2005 20:28
Picon

Re: Numeric + Lapack/Blas/ATLAS [Re: [BioPython] Installation help]

Catherine Letondal wrote:
 > So, could someone help? Has someone succedeed in the installation of
 > Numeric/Lapack/Atlas/Blas on MacOSX 10.3?  (The python version that is
 > used for the course is 2.3. )

Mac OS X comes with a preinstalled Lapack and Blas, so there is no need 
for Atlas. I am on travel this week and I can't check on my Mac right 
now, but what I remember is that I just needed to replace the 
libraries_list in Numeric's setup.py by ['LAPACK','BLAS'] (my apologies 
if my memory is not correct here).

> Another question/request is: would it be possible to use biopython 
> without Numeric?

Yes, but some packages will not work without it. For example, 
Bio.Cluster. You can have a look at biopython's setup.py to find out 
which other packages rely on Numeric.

--Michiel.
_______________________________________________
BioPython mailing list  -  BioPython <at> biopython.org
http://biopython.org/mailman/listinfo/biopython
Permalink | Reply | 0 comments |
headers
K. Arun | 10 Jan 2005 00:28
Picon

[BioPython] API documentation tarball ?

Hello,

Is the current API documentation available in compressed form
somewhere on the Biopython site ? I checked the usual places,
including the CVS repository without luck. Thanks,

-arun
_______________________________________________
BioPython mailing list  -  BioPython <at> biopython.org
http://biopython.org/mailman/listinfo/biopython
Permalink | Reply | 2 comments |
headers
Iddo Friedberg | 11 Jan 2005 00:17

Re: [BioPython] API documentation tarball ?

K. Arun wrote:

>Hello,
>
>Is the current API documentation available in compressed form
>somewhere on the Biopython site ? I checked the usual places,
>including the CVS repository without luck. Thanks,
>
>-arun
>_______________________________________________
>BioPython mailing list  -  BioPython <at> biopython.org
>http://biopython.org/mailman/listinfo/biopython
>
>
>  
>
You mean you would like a tarball of the HTML of the tutorial & cookbook 
for local reading? Or something else?

Cheers,

Iddo

--

-- 

Iddo Friedberg, Ph.D.
The Burnham Institute
10901 N. Torrey Pines Rd.
La Jolla, CA 92037
Tel: (858) 646 3100 x3516
(Continue reading)

Permalink | Reply | 0 comments |
headers
Iddo Friedberg | 11 Jan 2005 00:18

Re: [BioPython] API documentation tarball ?

K. Arun wrote:

>Hello,
>
>Is the current API documentation available in compressed form
>somewhere on the Biopython site ? I checked the usual places,
>including the CVS repository without luck. Thanks,
>
>-arun
>_______________________________________________
>BioPython mailing list  -  BioPython <at> biopython.org
>http://biopython.org/mailman/listinfo/biopython
>
>
>  
>

Oh sorry, you DID say the API. Let me see what I can do.

./I

--

-- 

Iddo Friedberg, Ph.D.
The Burnham Institute
10901 N. Torrey Pines Rd.
La Jolla, CA 92037
Tel: (858) 646 3100 x3516
Fax: (858) 713 9930
http://ffas.ljcrf.edu/~iddo
(Continue reading)

Permalink | Reply | 0 comments |
headers
Michiel Jan Laurens de Hoon | 11 Jan 2005 03:04
Picon

Re: [BioPython] numeric vs numarray

Ivan Rossi wrote:
> Any other evidence in favor or against numarray from people in the list?
> 
Few (if any) packages in Biopython critically depend on the speed of
numeric/numarray. For example, Bio.Cluster uses numeric but all
calculations are done internally in C (without using numeric's
functions), so I doubt that the speed of numeric / numarray will make a
difference. To minimize the hassle for users, I suggest that we stick to
numeric until other major packages have gone over, so users won't have
to install both numeric and numarray.

--Michiel.

_______________________________________________
BioPython mailing list  -  BioPython <at> biopython.org
http://biopython.org/mailman/listinfo/biopython
Permalink | Reply | 0 comments |
headers
paul m | 17 Jan 2005 18:45
Picon

[BioPython] Job Posting: Bioinformatics Programmer, Duke University (Durham, NC, USA)

Dear BioPythoneers,

My lab is looking to hire a bioinformatics programmer to participate
in research focused on comparative and functional genomics. I'm trying
to hire somebody local to the triangle area (Raleigh, Durham, Chapel
Hill) who can become part of the intellectual atmosphere of the lab.
The job description as posted to the Python jobs website follows
below.

Cheers,

Paul Magwene
Assistant Professor
Dept of Biology
Duke University

JOB DESCRIPTION:  Bioinformatics programmer sought for position in the
Department of Biology, Duke University. Be part of a team of
researchers using computational and experimental tools to study genome
evolution.  Python is our preferred development language, but the
person filling this position will also use C/C++ and/or Fortran to
implement numerically intensive algorithms.

QUALIFICATIONS: Bachelors or Masters degree in bioinformatics,
computer science, or molecular biology/evolution with 3+ years
programming experience.  Candidates must have demonstrated
capabilities in algorithm development and experience using
bioinformatics tools and databases (BLAST, Ensembl, etc).

* **Contact**: Dr. Paul Magwene
(Continue reading)

Permalink | Reply | 0 comments |

Gmane