Should maxquant tools fail when PXTQC fails?

[eschen42]

@npinter Whatcha think?

Both maxquant and maxquant_mqpar fail if PTXQC fails. I wonder whether this is appropriate.

In both cases, the command (optionally) ends with running the PTXQC report

    #if $qc.do_it:
        &&
        Rscript '$qr' '$qr_yaml'
        #if 'log' in $output:
            >> '$log' 2>&1
        #end if
        &&
        cp ./combined/txt/report_v@VERSION_PTXQC@_combined.pdf '$ptxqc_report'
    #end if

The trouble is that, even when MaxQuant completes successfully, if PTXQC fails then the MaxQuant results (obtained at great expense of time and computational cost) are not copied back from the cluster.

I would prefer to see something like this (writing freely here, not presenting something tested):

    #if $qc.do_it:
        && (
         (
          Rscript '$qr' '$qr_yaml'
          #if 'log' in $output:
              >> '$log' 2>&1
          #end if
          &&
          cp ./combined/txt/report_v@VERSION_PTXQC@_combined.pdf '$ptxqc_report'
         ) || (echo 'PTXQC report failed: see log')
       )
    #end if

This would allow the user (e.g., me) to salvage PTXQC failures by running PTXQC (which takes a few minute or less rather than several days) separately, outside of Galaxy.

[bernt-matthias]

My suggestion would be to put the R script in a separate tool.