WSSmod: Weighted Sum Scores for Protein Network Modules

Computes Weighted Sum Score (WSS) module-level summaries from proteomic data. Protein-level values are sign-aligned to their association with core biomarkers and combined within 'STRING' network modules, weighted by network connectivity and association strength. Bundled reference modules and a fitted 'joinet' model also allow all module scores to be predicted from age, sex and a small panel of plasma biomarkers, without measuring the full proteomic panel. The method is described in Butler et al. (2026) doi:10.64898/2026.08.14.26360394 .

Details

WSSmod computes Weighted Sum Score (WSS) values for protein network modules, and predicts them from a small panel of plasma biomarkers.

Scoring modules from proteomic data

• calculate_WSS() scores each module from a normalized and scaled samples-by-proteins matrix, using the bundled reference modules or your own.

• project_rank_norm() projects new values onto a frozen rank-normal reference, for normalizing new samples against a reference cohort.

Predicting module scores from plasma biomarkers

• normalize_wss_biomarkers() rank-normalizes raw biomarker values, for a single patient or a cohort.

• predict_WSS() predicts all module scores from age, sex and the normalized biomarkers (needs the suggested packages glmnet and joinet).

Bundled reference data

• list_prebuilt_wss(), load_prebuilt_wss(), load_prebuilt_terms(), load_prebuilt_wss_model() and load_prebuilt_biomarker_reference() give access to the prebuilt reference set.

• wss_example_expression and wss_example_biomarkers are small simulated datasets for trying out the functions.

The method is described in Butler et al. (2026), doi:10.64898/2026.08.14.26360394 .

See also

Useful links:

• https://github.com/Longo-Lab/WSSmod

• https://Longo-Lab.github.io/WSSmod/

• Report bugs at https://github.com/Longo-Lab/WSSmod/issues

Author

Maintainer: Robert Butler rrbutler@stanford.edu (ORCID)

Other contributors:

• WSSmod authors [copyright holder]