OMIX GSVA — Code Ocean Deployment Adapter

Calculate sample-level pathway enrichment scores from normalized continuous gene expression in Code Ocean while using the canonical OMIX GSVA scientific implementation unchanged.

Canonical OMIX module

Item	Location
Canonical module	OMIX GSVA
Interface contract	schemas/interface.yml
Development contract	OMIX module contract
Version and source record	OMIX_MODULE_SOURCE.md

The canonical module owns the method, defaults, and reusable interface. This repository owns only the Code Ocean App Panel, mounted-input discovery, /results placement, and runtime selection.

What this deployment adds

• Named App Panel controls for every user-settable canonical parameter.
• Recursive, content-based discovery of exactly one normalized-expression table and one sample-metadata table under /data when selectors are left blank; the default MSigDB compatibility database is attached internally.
• Stable Code Ocean results under /results.
• Selection of the imported OMIX Pathways v2 starter environment.

Inputs

Inputs may independently use CSV, TSV, or TXT delimiters. The adapter detects each file separately and provides advanced shared and per-input overrides. Supply files explicitly in the App Panel or attach assets containing exactly one unambiguous candidate for each role.

Normalized expression

A gene-by-sample table containing the selected gene column (default Gene) and numeric sample columns. Use normalized continuous expression—not raw integer counts.

Sample metadata

A table containing one row per sample and the selected sample-ID column (default Sample). Sample IDs must match expression column names.

Pathway database

A long table with these columns:

• collection
• gene_set_name
• gene_symbol
• species

The database is a separately provisioned, immutable Code Ocean Data Asset. It is not committed to this repository or embedded in the runtime. MSigDB content may be used only through an appropriately licensed restricted asset.

When a blank selector has no match or more than one match, the adapter stops and lists the candidate paths. It never chooses among ambiguous files.

Run the analysis

1. Attach one bundle containing normalized expression and matching sample metadata. In a workflow, replace the Expression and sample metadata input with an upstream result bundle. OMIX Limma Analysis may supply Batch_Corrected_Expression.csv and Sample_Metadata.csv for this role.
2. Confirm the gene and sample ID column names.
3. Use the attached MSigDB compatibility database, or explicitly select a different approved pathway database.
4. Select one or more exact pathway collection names.
5. Choose the GSVA method. The canonical default is gsva.
6. Run the capsule.

The App Panel defaults preserve the canonical module: Hallmark pathways, Human expression/database species, minimum set size 15, maximum set size 1200, and gene-symbol updating enabled.

Outputs and workflow handoff

Output	Purpose	Downstream use
gsva_results.csv	Gene-set-by-sample enrichment scores.	Continuous input to OMIX Limma Analysis.
gsva_heatmap.png	Row-scaled overview heatmap.	Visual review.
gsva_run_summary.txt	Inputs, selections, method, thresholds, and source provenance.	Reproducibility record.

Environment and reproducibility

• Runtime profile: r-pathway
• Code Ocean environment: codeocean/omix-r-pathway:r4.4.3-bioconductor3.20-v2
• Public immutable image: ghcr.io/nidap-community/omix-r-pathway@sha256:25ad9f50df5a4d7f70a7e3debed13f8d5c9b07cbc46d57fdf5c3cf4f2e767ddc
• Canonical source: recorded in OMIX_MODULE_SOURCE.md

Retain the Code Ocean computation ID, exact adapter commit, pathway Data Asset version, selected parameters, and released capsule version with every result.

Troubleshooting and limitations

• Raw counts are not supported. Normalize expression before GSVA.
• Ambiguous TXT delimiters require the corresponding advanced delimiter override.
• Collection names are exact and case-sensitive.
• Duplicate gene-set display names across selected collections are unsupported by the preserved implementation and should be resolved in the input asset.
• A Code Ocean run is still required before this adapter can be tagged or released. Local tests do not establish platform validation.

For developers

Read AGENTS.md and OMIX_MODULE_SOURCE.md before editing. code/functions/GSVA.R is a byte-identical canonical export and must not be edited in this repository.

References and support

• Hänzelmann S, Castelo R, Guinney J. BMC Bioinformatics. 2013;14:7.
• OMIX support: NCICCBRNIDAP@mail.nih.gov