← Files NGS Analysis WorkbenchARCHIVED FILE
workflows/scrnaseq_fastq_to_count/config/README.md
2.25 KB · Oct 5, 2026 · 18:12 UTC
# Configuration The package includes `config.json`, a runnable public single-cell example. URL reads and references are downloaded only when the workflow requires them. Local paths bypass downloading. No input data is included in the package. The Workbench uses this JSON config by default; supply an explicit JSON config to override it. Native Snakemake may also accept YAML, but the current MCP resolver reports non-JSON configs as `unknown` rather than guessing. The config contains: - `samples`: sample names mapped to barcode and cDNA FASTQ paths. - `threads`: default threads per rule. - `references`: genome FASTA, annotation GTF, and barcode whitelist paths. - `chemistry`: STARsolo barcode, UMI, filtering, and feature settings. - `chemistry.genome_sa_index_nbases`: STAR index sizing; the packaged chromosome-19 example uses 11, while configurations that omit the setting use STAR's whole-genome default of 14. - `chemistry.features_mode`: one supported gene-counting mode; raw matrices and, when cell filtering is enabled, filtered matrices use that same mode. - `execution.star_image`: the reviewed STAR image used when Snakemake deploys rules through Apptainer; Conda execution does not require an image. From the workflow package root, pass the packaged configuration explicitly: ```bash snakemake --snakefile workflow/Snakefile --configfile config/config.json --cores 4 ``` The workflow does not select a software deployment method automatically. The runtime may select the pinned `workflow/envs/star.yaml` environment for both STAR rules when Conda is available. Use the packaged container profile or an explicit deployment flag to select Apptainer instead: ```bash snakemake --snakefile workflow/Snakefile --configfile config/config.json --cores 4 --workflow-profile container snakemake --snakefile workflow/Snakefile --configfile config/config.json --cores 4 --sdm apptainer ``` When running Snakemake directly, pass the desired configuration explicitly. Input paths should be absolute when launched from a workbench run directory. The container image uses the standard `docker://` image format but runs through Apptainer, not a custom Docker invocation. The Workbench does not select Conda implicitly; the container profile can be selected when running Snakemake directly.
SHA-256: dc379e3f1864aa69e0f63bb21e2f339dfe99fbd722591a523e2cf2a1128099ae