Code to reproduce results from Heiser, et al. (2023).
These analyses integrate single-cell RNA sequencing (scRNA-seq), spatial transcriptomics (ST), and multiregional exome sequencing via laser-capture microdissection (LCM-WES) to generate models and metrics presented in our Cell manuscript, "Molecular cartography uncovers evolutionary and microenvironmental dynamics in sporadic colorectal tumors". Functions herein include:
- Reading in and pre-processing data
- Scoring literature-derived gene signatures in scRNA-seq and ST
- Somatic CNV inference in scRNA-seq and ST
- Phylogenetic analysis from LCM-WES
- refNMF cell state deconvolution between scRNA-seq and ST
- MILWRM tissue domain detection
- Summaries and visualizations of integrated spatial data
- GAM modeling of genes, gene signatures, and cell states along progression pseudotime (PPT)
- Immune Exclusion analysis in ST
Annotated, pre-processed data compatible with the codebase can be downloaded from OSF Storage: https://osf.io/hftq2/
All raw data have been deposited to the HTAN Data Coordinating Center Data Portal at the National Cancer Institute: https://data.humantumoratlas.org/ (under the HTAN Vanderbilt Atlas).
Code in this repository is intended to be run in order, as some analyses depend on outputs from previous tools. We therefore divide up the analysis notebooks into steps, provided as ordered directories.
- Download publicly available data from OSF to the
data/directory first - Proceed to the notebooks in the
step1/directory, thenstep2/, andstep3/and so forth. All notebooks residing in the same directory can be run in any order, but they all must be successfully run before proceeding to the next step.
