TractorWorkflow

Requires Tractor >= v2.6.0. TractorWorkflow is versioned independently of Tractor; this release brings the workflow in sync with the Tractor v2.6.0 command-line interface. Get Tractor v2.6.0 here. Multi-phenotype Tractor GWAS in a single task per chromosome. --phenolist_file now runs all listed phenotypes in one task instead of one task per phenotype, thus reading the hapcount/dosage files only once and improving on I/O. One summary statistics file is published per phenotype, suffixed _ . New Tractor arguments exposed. --progress_every for periodic progress reporting on long runs. --logistic_backend to select fastglm (default) or glmfit. Improved sample-exclusion reporting. phenotype_samples_not_in_hapdose_files_ *.txt and hapdose_samples_not_in_phenotype_file_ *.txt replace samples_excluded_from_phenotype.txt, and are now tagged by chromosome so per-chromosome reports no longer overwrite each other. Breaking changes --totallines has been removed. Passing it now raises an explicit error rather than failing silently. Requires Tractor v2.6.0 or later. Earlier Tractor versions do not accept --progress_every or --logistic_backend. fastglm must be installed in the Tractor conda environment for --regression_method logistic. Tractor v2.6.0 renames the logistic test statistic from tval_anc* to zval_anc* and suffixes output files with the phenotype name. See the Tractor CHANGELOG before re-running existing analyses. Fixed Relative paths for --phenotype no longer fail.