Configuration¶

Three configuration layers are resolved for each run:

1. packaged workflow defaults define assay behavior
2. site configuration defines cluster resources and scheduler behavior
3. the generated run configuration records resolved CLI and runtime settings

Run configurations are written to EXPERIMENT_DIR/run_configs/ and should be retained with the analysis provenance.

User site configuration¶

The packaged template is src/omnomnomics/workflow/config/site.yaml. Copy it to:

$XDG_CONFIG_HOME/omnomnomics/site.yaml

or, when XDG_CONFIG_HOME is unset:

~/.config/omnomnomics/site.yaml

Key settings include:

partition: "rome"
max_jobs: 100
max_jobs_per_second: 2
max_status_checks_per_second: 2

controller_partition: "rome"
controller_cores: 1
controller_mem_mb: 8192
controller_runtime: 1440

worker_constraint: null
heavy_io_constraint: null
genome_assembly_dir: "{HOME}/genomes/assemblies"

Cluster feature names are site-specific. Set worker_constraint only when all worker jobs require that feature; otherwise leave it empty. Use --site-config for a one-run override without modifying the persistent user file.