Hi Tin, On Fri, Aug 12, 2016 at 01:56:39PM +0000, Duy Tin Truong wrote: > Yes, that is fine for me. Just a small comment, metaphlan2-data will depend > on bowtie2 so that you can have bowtie2 to convert the fasta file.
Yes, that's obvious. > In addition, this is the list of dependencies for metaphlan2 in case you > need: > https://bitbucket.org/biobakery/metaphlan2#markdown-header-pre-requisites Most of them are available in Debian and the current packaging says: Depends: python-biom-format, python-msgpack, python-pandas, bowtie2 (while python-biom-format implicitly depends python-numpy and python-scipy) However the hint ot hclust2[1] is helpful. Unfortunately I can not find any description for this software. Since you might have some influence on this it would be great to provide a hint where I can find a description for a potential package. > and here for strainphlan (another brother tool uses the same database with > metaphlan2 and both are in the same repository and should go together, > strainphlan is in strainphlan.py and metaphlan2 is in metaphlan2.py): > https://bitbucket.org/biobakery/metaphlan2#markdown-header-pre-requisites_1 Well, the download file for metaphlan2 version 2.5 has strainer_src and metaphlan2_strainer.py - is this what you mean? Kind regards Andreas. [1] https://bitbucket.org/nsegata/hclust2 > > On Fri, Aug 12, 2016 at 1:59 PM Andreas Tille <[email protected]> wrote: > > > Hi Duy, > > > > On Thu, Aug 11, 2016 at 05:03:20AM +0000, Duy Tin Truong wrote: > > > We have discussed the plan but unfortunately, we do not have enough > > > resource for that task now. In addition, redesigning the source structure > > > requires us to change all tutorials and that is quite expensive for us > > and > > > users now. We will inform you when we can separate them. Currently, the > > > tutorial for users is here: > > > https://bitbucket.org/biobakery/metaphlan2 > > > > > > and fits well with latest version and it is quite stable now. We don't > > > think that there will be a substantial update for the source code in the > > > near feature. > > > > Thanks for the explanation which helps me to make a sensible decision. > > I plan to do the following: > > > > 1. metaphlan2-data > > The source tarball will be created by downloading the original > > tarball from your site, strip the code and convert the data using > > bowtie2-build markers.fasta ../db_v20/mpa_v20_m200 > > The Debian source tarball created this way will ship the fasta > > version of the data and rebuilds the bowtie2 database at > > installation time on users machine. I plan to enable the admin > > to opt out from immediate generation and provide a script that > > does the job later. I also plan to provide md5sums of the data > > to ensure that the resulting database is really identical to the > > metaphlan2 download. > > > > 2. metaphlan2 (the code): > > The source tarball will be created by simply striping the data > > and the binary Debian package as its done currently in my > > packaging code. The metaphlan2 package will depend from the > > metaphlan2-data package > > > > This does not require any change at your side but prevents over-large > > packages on Debian site. > > > > Please confirm that this plan sounds sensible to you (or if I was not > > explicite enough in my explanation). > > > > Kind regards > > > > Andreas. > > > > -- > > http://fam-tille.de > > -- http://fam-tille.de
