Re: About pysam
Hello,
I made an effort to try and clean up the code duplication mess of pysam /
samtools. My first attempt was to provide upstream with a suggestion on how to
build tabix as a shared library.[1] (As a step toward convincing them to make
samtools build libbam as a shared library)
However I don't think they took advantage of it.
I then discovered a worse code duplication problem in pysam & samtools with
the attractivechaos/klib "library" [2] as far as I can tell a number of
bioinformatics packages just periodically copy the source from that repository
into their own projects.
At that point I gave up, and just built a package for my lab.
I recall seeing something about an improved libbam or samtools? (I can't
quickly find Charles' posts on the list about it though)
I could try cleaning the package up a bit for Debian, but I don't think I
could resolve the code duplication issues. (I could at least get the copyright
/ README.Debian to acknowledge and point to where all the code is being
duplicated from).
Diane
[1] my patch https://github.com/detrout/tabix/tree/dynamic-makefile
[2] https://github.com/attractivechaos/klib
On Tuesday, February 11, 2014 09:08:14 Andreas Tille wrote:
> Hi Charles,
>
> thanks for your comments. Just an explanation for my engagement. I'd like
> to package
>
> http://blends.debian.org/med/tasks/bio#gasic
> http://blends.debian.org/med/tasks/bio#fitgpc
>
> and my reasons are given in the according remarks there.
>
> On Tue, Feb 11, 2014 at 01:43:16PM +0900, Charles Plessy wrote:
> > Hi Andreas,
> >
> > I see that you are working on pysam, thank you for this !
> >
> > If you are wondering why it was never uploaded, it is because of the
> > extensive code duplications with the samtools and tabix packages, which I
> > could not try to resolve because of my lack of understanding on how the C
> > code is used in python (in particular the relation between original code
> > and the pxd or pyx files).
>
> This became obvious in the according remarks you have given in debian/DRAFT.
> > See also https://lists.debian.org/debian-med/2012/12/msg00112.html, it
> > looks like Diane prepared a patch.
>
> I can confirm that I checked out Diane's repository (Diane in CC) but I
> do not see any patch for dealing with the code duplication. Diane, am I
> missing something?
>
> > Maybe I was overzealous, so if you feel like uploading, I definitely do
> > not
> > mind ! I am using Debian and pysam in one of my tutorials for
> > transcriptome analysis
> > (https://github.com/charles-plessy/tutorial/search?q=pysam&ref=cmdform).
> My main goal was to get all tests working first. Once the tests are
> properly working I could try to get rid of code copies and if I manage
> to reproduce the tests this could be considered successful. I will keep
> you updated about the progress.
>
> Thanks for your input
>
> Andreas.
Reply to: