[Debian-med-packaging] Bug#458926: Bug#458926: probcons contains incorrect probcons-RNA sub-program

[Charles Plessy]

Le Thu, Jan 03, 2008 at 01:01:45PM -0500, Benjamin Redelings a écrit :
> 
> The probcons package contains a binary probcons-RNA that is very 
> different than the one that is provided at this page:
> 
>    http://probcons.stanford.edu/download.html
> 
> It gives very different result when applied to a collection of diverse 
> 5S RNAs, producing a huge alignment that is mostly gaps.  Attempts to 
> train it (using the -t option) produce a parameter file that contains an 
> amino-acid transition matrix, NOT a nucleic-acid transition matrix.

Dear Benjamin, many thanks for the report.

Apparently something is going wrong in the building of probconsRNA. The version
provided upstream is 1.10, whereas probcons iteslf is now 1.12.  In Debian, we
build probconsRNA from probcons 1.12. I attached the the difference between
probcons 1.10 and probconsRNA 1.10. As you can see, only Defaults.h and the
Makefile differ.  Therefore, instead of using the original sources of
probconsRNA, we ship Defaults.h, renamed DefaultsRNA.h, copy Main.cc as
MainRNA.cc, and build probconsRNA with the rest of the probcons sources version
1.12. (see the following patch for the details:)

http://svn.debian.org/wsvn/debian-med/trunk/packages/probcons/trunk/debian/patches/10_add-RNA-in-Makefile.diff?op=file&rev=0&sc=0

Can you provide me your sequence files so that we have a common
reference ?

Have a nice day,

-- 
Charles Plessy
Debian-med packaging team.
Wakō, Saitama, Japan
-------------- next part --------------
A non-text attachment was scrubbed...
Name: probconsRNA.diff
Type: text/x-diff
Size: 8355 bytes
Desc: not available
Url : http://lists.alioth.debian.org/pipermail/debian-med-packaging/attachments/20080104/f251eca5/attachment.diff