On 10/25/2012 10:54 AM, Peter Cock wrote:
On Thu, Oct 25, 2012 at 9:36 AM, Jorrit Boekel <jorrit.boekel@scilifelab.se> wrote:
Dear list,
In my galaxy fork, I extensively use the job splitters. Sometimes though, I have to split to different file types for the same job. That raises an exception in the lib/galaxy/jobs/splitters/multi.py module.
I have turned this behaviour off for my own work, but am now wondering whether this is very bad practice. In other words, does somebody know why the multi splitter does not support multiple file type splitting?
cheers, jorrit Could you clarify what you mean by showing some of your tool's XML file. i.e. How is the input and its splitting defined.
Are you asking about splitting two input files at the same time?
Peter
Hi Peter, Something like the following: <command interpreter="python">bullseye.py $hardklor_results $ms2_in.extension $ms2_in $output $use_nonmatch</command> <parallelism method="multi" split_inputs="hardklor_results,ms2_in" shared_inputs="config_file" split_mode="from_composite" merge_outputs="output"/> <inputs> The tool takes two datasets of different formats, which are to be split in the same amount of files, which belong together as pairs. Note that I have implemented an odd way of splitting, which is from a number of files in the dataset.extra_files_path to symlinks in the task working dirs. The number of files is thus equal to the number of parts resulting from a split, and I have ensured that each part is paired correctly. I assume this hasn't been necessary in the genomics field, but for proteomics, at least in our lab, multiple-file datasets are the standard. My fork is at http://bitbucket.org/glormph/adapt if you want to check more closely. cheers, jorrit -- Scientific programmer Mass spec analysis support @ BILS Janne Lehtiö / Lukas Käll labs SciLifeLab Stockholm