Code scale-up, parallelisation, etc etc
Short recap down there of the difft options to get cc-qtl-workflow to proceed faster:
qtl2-specific arguments:
- cores argument: add "cores = cores" in qtl2 functions that allow it, def cores as arg of Rscript ; and in the xml wrapper Rémi takes care of VARENV that grabs defined nb of cores on galaxy instance.
- max_batch (for scan1): indicates the maximum number of phenotypes to run together; default is unlimited (hence can trigger memory pb...). Back of envelope calculation ~ upper limit to 500 pheno being analyzed at once. Phenos should apparently be considered one at a time iff LMM and not HK (https://groups.google.com/g/rqtl2-disc/c/M9clhYsM0LU for discussion)
Others: permutation procedure ? to be thought about.