Analysis code for studying disease-modifying antirheumatic drug (DMARD) treatment response in rheumatoid arthritis using single-cell RNA sequencing of pretreatment peripheral blood mononuclear cells (PBMCs). The analyses compare responders (R) and nonresponders (NR).
- Responder Score: cytokine-reference-based scoring and comparison of response groups.
- Network similarity: comparison of patient-specific functional gene networks within cell types.
- scGPT in silico perturbation: simulated gene-expression perturbations and evaluation of their effects in IFN CM cells.
Each analysis README describes its methods, inputs, and execution steps.
RA-DMARD-response/
├── README.md
├── .gitignore
├── 01_cytokine_reference_responder_score/
├── 02_network_similarity/
└── 03_scGPT_perturbation/
Raw and processed single-cell data are provided through the Gene Expression Omnibus (GEO) and are not included in this repository. Patient-identifiable information and private clinical data are not distributed here.
Environments are analysis-specific. Each analysis directory provides a requirements.txt for conda. The scGPT directory additionally provides requirements-pip.txt and requirements-r.txt for its Python dependencies and separate R environment.
Follow the installation instructions in the corresponding analysis README, including the additional scHumanNet/ACTIONet/SCINET and scGPT/flash-attention setup where applicable. Network construction and Seurat input preparation also use separate environments.
Clone the repository and enter the analysis directory of interest. For example, to run the Responder Score workflow, replace <repository-url> with this repository's clone URL:
git clone <repository-url> RA-DMARD-response
cd RA-DMARD-response/01_cytokine_reference_responder_score
cp config.example.yaml config.yamlRead the analysis README, prepare its environment and required inputs, then edit config.yaml and follow the documented execution order.
Each analysis records major parameters and input/output paths in config.example.yaml; package requirements and numerical tests are provided alongside the code. Large intermediate files, model checkpoints, and generated outputs are excluded from GitHub through .gitignore.