Skip to content

[Bioc-devel] Dependency to PEER

5 messages · Kasper Daniel Hansen, Robert Castelo, Christian Mertes +1 more

Christian Mertes
# 
Hi all,

Im the developer of OUTRIDER. Our package normalises RNA-seq count data and finds expression outliers within the population.
During the development we compared our approach against PCA and PEER. The later one is a well established method to normalise RNA-seq data.

https://github.com/gagneurlab/OUTRIDER <https://github.com/gagneurlab/OUTRIDER>
https://github.com/PMBio/peer <https://github.com/PMBio/peer>

For PEER, an R package exists on GitHub, but it is not in CRAN nor Bioconductor, even it would fit really well with Bioconductor. 

We would like to leave the user the option to switch between the different normalisation methods within our package (auto encoder, peer, pca, sea, ?). But since the Bioconductor requirements are that all loaded libraries need to be within CRAN or BioC, my question is how we should implement this.

Is there a best practice for this? Does we have to ask the developer of PEER to push it on CRAN or do we get around this? Or can we flag one function to be not checked by Bioconductor for dependencies and use require(peer) within the function?

Thanks for any advices in advance!

Best,
Christian
Kasper Daniel Hansen
# 
It is a major weakness of PEER that it has not been submitted to a standard
repository. Of course, that is up to the PEER developers, but the community
should put pressure on them.
On Tue, Sep 25, 2018 at 12:40 PM Christian Mertes <mertes at in.tum.de> wrote:

            

      
      
      
    

        

      
      
    

        

  
  


  


      

    

    

      
      

        


  

        

      Robert Castelo
    

    

      
      #
    

  


  

      
hi,

it seems that the PEER repository hasn't been updated in the last 7 
years and that the package is distributed under a GPL-2 license. if i'm 
interpreting this correctly, you could copy&paste that code into your 
package redistributing it under the GPL-2 license and asking the users 
of PEER to cite the PEER paper when using it.

cheers,

robert.

        
On 25/09/2018 18:55, Kasper Daniel Hansen wrote:

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Christian Mertes
    

    

      
      #
    

  


  

      
Hi,

thanks for the answers. 

Since we only use 5 functions from the package within one function in our package, I hoped to get around it by masking this function in a special way.

The package is not dependent on PEER and works without it perfectly. Its just to make it easier for the user if he wants to use a different normalisation method.

Best,
Christian

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

            

      
      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Martin Morgan
    

    

      
      #
    

  


  

      
On 09/25/2018 07:07 PM, Christian Mertes wrote:

            

      
      
      
    

        
not sure what that means; better not to be devious if that is what 
'special' means.

I think an appropriate thing to do would be to illustrate installation / 
evaluation functionality in a non-evaluated code chunk in the vignette. 
Your reviewer will be understanding.

Martin