Code to reproduce methods & results from Boiarsky et al., Nature Communications, 2022

If you use our data or analysis in your research, please cite our research article!

Boiarsky, R., Haradhvala, N.J., Alberge, JB. et al. Single cell characterization of myeloma and its precursor 
conditions reveals transcriptional signatures of early tumorigenesis. Nat Commun 13, 7040 (2022). 
https://doi.org/10.1038/s41467-022-33944-z

Our code was run using the following software:

• Python version 3.8.13
• R version 4.1.0

• scanpy==1.7.1
• datatable
• scikit_posthocs
• statsmodels
• scipy
• seaborn
• re

other:

• clone the SignatureAnalyzer-GPU repo to run Signature Analyzer NMF decomposition

• stringr_1.4.0
• tibble_3.1.7
• Matrix_1.4-1
• edgeR_3.36.0
• ggplot2_3.3.6
• tidyr_1.2.0
• dplyr_1.0.9
• limma_3.50.3

To create an anndata object from the raw data publicly available on GEO (accession number GSE193531) and the supp/source tables published with the paper, follow the code in the notebook 0_reproduce_results_from_raw_data.ipynb (this notebook is being updated, further sections will be completed soon).

The notebook 0_reproduce_results_from_raw_data.ipynb is a good starting place if one wants to explore our data themselves. More detailed versions of our analysis code is contained in the following notebooks:

• Ig_genes.ipynb details how we determined which genes fall in the immunoglobulin loci, to remove them from downstream analyses.
• 4a_puritywork-published.ipynb contains our analysis of sample purity (% tumor cells in sample) using our Bayesian purity model. It also contains the code to generate Fig. 2a from our paper.
• 4d_limma.ipynb contains our limma-voom differential expression analysis comparing malignant or pre-malignant pseudobulk samples vs. normal pseudobulk samples.
• 5_NMF_rawdata-moreHVG-published.ipynb contains code related to generating the input data for SignatureAnalyzer and our analysis of SignatureAnalyzer results, including Figs. 3a-d from our paper.
• 5b_heterogeneity.ipynb contains code for analyzing the heterogeneity of signature expression within tumor samples, including Fig. 4c from our paper.
• helper_functions_published.py contains functions that are used throughout the other notebooks included in the repo.