[PATCH] Add Ribotaper.

[PATCH] Add Ribotaper.

[Ricardo Wurmus]

Hi Guix,

this patch adds a bioinformatics tool called "Ribotaper".  It needs a somewhat
older version of bedtools, because the output format produced by bedtools
changed after version 2.18.0 (they don't seem to care about semantic
versioning).  This is why this patch not only adds a variable "ribotaper" but
also "bedtools-for-ribotaper".

Do you think I should rather name it "bedtools-2.18" and make it public so
that users can install it?  Or is it better to keep it private and name it
"bedtools-for-ribotaper"?

~~ Ricardo

Ricardo Wurmus (1):
  gnu: Add Ribotaper.

 gnu/packages/bioinformatics.scm | 44 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)

-- 
2.8.4

[PATCH] gnu: Add Ribotaper.

[Ricardo Wurmus]

* gnu/packages/bioinformatics.scm (bedtools-for-ribotaper, ribotaper):
  New variables.
---
 gnu/packages/bioinformatics.scm | 44 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)

diff --git a/gnu/packages/bioinformatics.scm b/gnu/packages/bioinformatics.scm
index 294f4ee..c4af842 100644
--- a/gnu/packages/bioinformatics.scm
+++ b/gnu/packages/bioinformatics.scm
@@ -258,6 +258,50 @@ intervals from multiple files in widely-used genomic file formats such as BAM,
 BED, GFF/GTF, VCF.")
     (license license:gpl2)))
 
+;; Later releases of bedtools produce files with more columns than
+;; what Ribotaper expects.
+(define bedtools-for-ribotaper
+  (package (inherit bedtools)
+    (name "bedtools")
+    (version "2.18.0")
+    (source (origin
+              (method url-fetch)
+              (uri (string-append "https://github.com/arq5x/bedtools2/"
+                                  "archive/v" version ".tar.gz"))
+              (file-name (string-append name "-" version ".tar.gz"))
+              (sha256
+               (base32
+                "05vrnr8yp7swfagshzpgqmzk1blnwnq8pq5pckzi1m26w98d63vf"))))))
+
+(define-public ribotaper
+  (package
+    (name "ribotaper")
+    (version "1.3.1")
+    (source (origin
+              (method url-fetch)
+              (uri (string-append "https://ohlerlab.mdc-berlin.de/"
+                                  "files/RiboTaper/RiboTaper_Version_"
+                                  version ".tar.gz"))
+              (sha256
+               (base32
+                "0ykjbps1y3z3085q94npw8i9x5gldc6shy8vlc08v76zljsm07hv"))))
+    (build-system gnu-build-system)
+    (inputs
+     `(("bedtools" ,bedtools-for-ribotaper)
+       ("samtools" ,samtools-0.1)
+       ("r" ,r)
+       ("r-foreach" ,r-foreach)
+       ("r-xnomial" ,r-xnomial)
+       ("r-domc" ,r-domc)
+       ("r-multitaper" ,r-multitaper)
+       ("r-seqinr" ,r-seqinr)))
+    (home-page "https://ohlerlab.mdc-berlin.de/software/RiboTaper_126/")
+    (synopsis "Define translated ORFs using ribosome profiling data")
+    (description
+     "RiboTaper is a method for defining translated ORFs using ribosome
+profiling data.")
+    (license license:gpl3+)))
+
 (define-public bioawk
   (package
     (name "bioawk")
-- 
2.8.4

Re: [PATCH] Add Ribotaper.

[Ben Woodcroft]

On 07/07/16 19:26, Ricardo Wurmus wrote:
> Hi Guix,
>
> this patch adds a bioinformatics tool called "Ribotaper".  It needs a somewhat
> older version of bedtools, because the output format produced by bedtools
> changed after version 2.18.0 (they don't seem to care about semantic
> versioning).  This is why this patch not only adds a variable "ribotaper" but
> also "bedtools-for-ribotaper".
>
> Do you think I should rather name it "bedtools-2.18" and make it public so
> that users can install it?

This would be my preference, IIUC this is the current way we do this. Is 
there any difference from previous times we've had to include outdated 
packages?

Note that I just updated bedtools, although I imagine that won't help here.

Thanks.
ben

Re: [PATCH] Add Ribotaper.

[Ricardo Wurmus]

Ben Woodcroft <b.woodcroft@uq.edu.au> writes:

> On 07/07/16 19:26, Ricardo Wurmus wrote:
>> Hi Guix,
>>
>> this patch adds a bioinformatics tool called "Ribotaper".  It needs a somewhat
>> older version of bedtools, because the output format produced by bedtools
>> changed after version 2.18.0 (they don't seem to care about semantic
>> versioning).  This is why this patch not only adds a variable "ribotaper" but
>> also "bedtools-for-ribotaper".
>>
>> Do you think I should rather name it "bedtools-2.18" and make it public so
>> that users can install it?
>
> This would be my preference, IIUC this is the current way we do this. Is 
> there any difference from previous times we've had to include outdated 
> packages?

I actually prefer to just offer “bedtools-2.18” without restricting it
to just ribotaper, but we’d have to ensure that this version stays
around (adding a comment on top should be enough).

We currently have “guile-for-guile-emacs”, which offers a variant of
Guile, but the situation is slightly different here.  We also have
“armadillo-for-rcpparmadillo”.  In both cases, the packages are not
private.  In the former case the package name differs, but in the case
of armadillo it’s only the variable name that indicates that this is
special.

I think in the ribotaper/bedtools case we should just name the variable
“bedtools-2.18”  and make it public.

> Note that I just updated bedtools, although I imagine that won't help here.

No, the format hasn’t changed back.  Ribotaper was not developed with
the more recent versions, which all output more columns.

~~ Ricardo

Re: [PATCH] Add Ribotaper.

[Ben Woodcroft]

On 07/07/16 23:45, Ricardo Wurmus wrote:
> Ben Woodcroft <b.woodcroft@uq.edu.au> writes:
>
>> On 07/07/16 19:26, Ricardo Wurmus wrote:
>>> Hi Guix,
>>>
>>> this patch adds a bioinformatics tool called "Ribotaper".  It needs a somewhat
>>> older version of bedtools, because the output format produced by bedtools
>>> changed after version 2.18.0 (they don't seem to care about semantic
>>> versioning).  This is why this patch not only adds a variable "ribotaper" but
>>> also "bedtools-for-ribotaper".
>>>
>>> Do you think I should rather name it "bedtools-2.18" and make it public so
>>> that users can install it?
>> This would be my preference, IIUC this is the current way we do this. Is
>> there any difference from previous times we've had to include outdated
>> packages?
> I actually prefer to just offer “bedtools-2.18” without restricting it
> to just ribotaper, but we’d have to ensure that this version stays
> around (adding a comment on top should be enough).
>
> We currently have “guile-for-guile-emacs”, which offers a variant of
> Guile, but the situation is slightly different here.  We also have
> “armadillo-for-rcpparmadillo”.  In both cases, the packages are not
> private.  In the former case the package name differs, but in the case
> of armadillo it’s only the variable name that indicates that this is
> special.
>
> I think in the ribotaper/bedtools case we should just name the variable
> “bedtools-2.18”  and make it public.

I agree with should call it 'bedtools-2.18' and make it public (but with 
a comment explaining why the old version is kept), just as in e.g. 
'python-flake8-2.2.4'.

Are you suggesting we keep bedtools-2.18 even if ribotaper updates to a 
newer version? I thought accepted practice was to delete old packages 
when they are no longer needed.

The patch LGTM otherwise, except that I would unpack 'ORF' and use the 
term 'ribo-seq' in the description. I've never looking at ribo-seq data 
before and I'm not familiar, so I wasn't sure what 'ribosome profiling' 
meant exactly e.g. is it measuring the amount of ribosome in a cell? I 
realise that isn't true, just illustrating the point. Perhaps I am too 
naive.

ben

Re: [PATCH] Add Ribotaper.

[Ricardo Wurmus]

Ben Woodcroft <b.woodcroft@uq.edu.au> writes:

>> I think in the ribotaper/bedtools case we should just name the variable
>> “bedtools-2.18”  and make it public.
>
> I agree with should call it 'bedtools-2.18' and make it public (but with 
> a comment explaining why the old version is kept), just as in e.g. 
> 'python-flake8-2.2.4'.
>
> Are you suggesting we keep bedtools-2.18 even if ribotaper updates to a 
> newer version? I thought accepted practice was to delete old packages 
> when they are no longer needed.

No, I agree.  We keep bedtools-2.18 as long as it’s needed for
Ribotaper.

> The patch LGTM otherwise, except that I would unpack 'ORF' and use the 
> term 'ribo-seq' in the description. I've never looking at ribo-seq data 
> before and I'm not familiar, so I wasn't sure what 'ribosome profiling' 
> meant exactly e.g. is it measuring the amount of ribosome in a cell? I 
> realise that isn't true, just illustrating the point. Perhaps I am too 
> naive.

I don’t know how to naturally include the term “ribo-seq”, so how about
simply this:

   Ribotaper is a method for defining translated @dfn{open reading
   frames} (ORFs) using ribosome profiling (ribo-seq) data.  This
   package provides the Ribotaper pipeline.

Is this better?

~~ Ricardo

Re: [PATCH] Add Ribotaper.

[Ben Woodcroft]

On 10/07/16 05:43, Ricardo Wurmus wrote:
> Ben Woodcroft <b.woodcroft@uq.edu.au> writes:
>
>
>> The patch LGTM otherwise, except that I would unpack 'ORF' and use the
>> term 'ribo-seq' in the description. I've never looking at ribo-seq data
>> before and I'm not familiar, so I wasn't sure what 'ribosome profiling'
>> meant exactly e.g. is it measuring the amount of ribosome in a cell? I
>> realise that isn't true, just illustrating the point. Perhaps I am too
>> naive.
> I don’t know how to naturally include the term “ribo-seq”, so how about
> simply this:
>
>     Ribotaper is a method for defining translated @dfn{open reading
>     frames} (ORFs) using ribosome profiling (ribo-seq) data.  This
>     package provides the Ribotaper pipeline.
>
> Is this better?

Thanks for indulging my naivety. Looks good.

ben