Case Study
This page presents the scripts used for the manuscript case study applying the SimiC-Suite workflow to the Boiarsky et al. NBM, SMM, and MM plasma-cell scRNA-seq dataset. The scripts are included as static resources for transparency and reproducibility.
Dataset
This case study applies SimiC-Suite to the single-cell RNA-seq cohort from Boiarsky et al. (2022), which profiles CD138+ plasma cells across normal bone marrow and multiple stages of myeloma progression.
The original dataset contains four disease labels: normal bone marrow (NBM), monoclonal gammopathy of undetermined significance (MGUS), smoldering multiple myeloma (SMM), and multiple myeloma (MM). For this analysis, we retained the NBM, SMM, and MM groups and excluded MGUS samples.
MGUS was excluded because it contained relatively few cells and was reported to have a low estimated fraction of malignant plasma cells. Consequently, the MGUS population was expected to contain a substantial proportion of normal plasma cells, making it less representative of a distinct malignant disease state and potentially confounding phenotype-specific regulatory network inference.
Workflow Overview
The workflow follows the same overall structure as the SimiC-Suite manuscript case study:
- Retrieve and prepare the Boiarsky scRNA-seq dataset.
- Run SimiCPipeline preprocessing to generate imputed expression, transcription factor lists, selected target genes, and phenotype annotations.
- Run SimiCPipeline for phenotype-specific GRN inference and RAS calculation.
- Use SimiCviz to inspect GRN quality, compare regulatory programs, and generate downstream figures.
- Project RAS values onto the processed AnnData UMAP for visual interpretation.
The public GEO matrix is used directly.
Requirements
Python requirements include simicpipeline, anndata, pandas, numpy, scipy, scikit-learn, scanpy, magic-impute, and scprep.
R requirements include SimiCviz, ComplexHeatmap, circlize, viridisLite, dplyr, tidyr, tibble, and grid.
Workflow Scripts
01 Data Retrieval
Python script for downloading and preparing the Boiarsky et al. dataset, filtering stages, generating QC plots, and exporting SimiCPipeline-compatible inputs.
02 SimiCPipeline Preprocessing
Jupyter notebook for SimiCPipeline preprocessing, including imputation, transcription factor selection, target gene selection, and phenotype annotation setup.
03 SimiCPipeline Run
Python script for running SimiCPipeline cross-validation, GRN inference, filtering, and RAS calculation.
04 SimiCviz Explorations
R / SimiCviz script for loading SimiCPipeline outputs, exploring GRN quality, calculating dissimilarity, and generating downstream case-study figures.
05 RAS on UMAP
Python script for overlaying RAS values from the SimiCPipeline output onto the processed AnnData UMAP.
Dataset Citation
Boiarsky, R., Haradhvala, N.J., Alberge, J.B. et al. Single cell characterization of myeloma and its precursor conditions reveals transcriptional signatures of early tumorigenesis. Nature Communications 13, 7040 (2022). https://doi.org/10.1038/s41467-022-33944-z
Please also cite the SimiC-Suite manuscript when using this workflow.