This repository contains the R package for the Spatial Topic Model we have developed for flexible and interpretable analysis of spatial transcriptomics data. The method integrates multiple topic modelling frameworks to enable:
- Inferrence of transcriptional programs that can predict histology-based annotations
- Identification of spatially informed clusters through neighbor prediction tasks
- Cell type Deconvolution (with or without a reference single cell dataset)
- Impute spatial information in single cell atlases
Our manuscript (now published in Briefings in Bioinformatics) detailing the method's performance and analyses can be found here. The analysis scripts used to generate the results found in the manuscript (using version 0.3 of SpaTM) can be found here.
devtools::install_github("li-lab-mcgill/SpaTM")
Please use the issues tab to contact us if you encounter any issues/bugs or have suggestions for our software!
