I think we have the basic features ready in this pipeline.
@ypriverol @timosachsenberg
Unfortunately most bug reports that we receive are from people struggling with SDRF or getting errors in the middle of the pipeline because of erroneous SDRFs.
I think we should significantly extend and improve both the checks and error messages in multiple steps, starting with SDRF parsing before focusing on even more complicated designs such as TMT with MBR. We can add erroneous SDRFs to the sdrf-pipelines repository and make sure that reasonable error messages are given. We can add several validation levels: for openms, for quantms, for quantms with MSstats (or downstream statistics), with each level requiring more information. We maybe should even invest in a basic UI component to more easily complete an SDRF in a way that is suitable for this pipeline. We can build on lesSDRF.
Other steps that would benefit from error handling are ID and filtering steps, where we could double-check that ID files are not empty.
Maintainers should also make sure to forward bad error handling to the upstream tools such as OpenMS.
I think we have the basic features ready in this pipeline.
@ypriverol @timosachsenberg
Unfortunately most bug reports that we receive are from people struggling with SDRF or getting errors in the middle of the pipeline because of erroneous SDRFs. I think we should significantly extend and improve both the checks and error messages in multiple steps, starting with SDRF parsing before focusing on even more complicated designs such as TMT with MBR. We can add erroneous SDRFs to the sdrf-pipelines repository and make sure that reasonable error messages are given. We can add several validation levels: for openms, for quantms, for quantms with MSstats (or downstream statistics), with each level requiring more information. We maybe should even invest in a basic UI component to more easily complete an SDRF in a way that is suitable for this pipeline. We can build on lesSDRF.
Other steps that would benefit from error handling are ID and filtering steps, where we could double-check that ID files are not empty.
Maintainers should also make sure to forward bad error handling to the upstream tools such as OpenMS.