Our pipeline, SPoTLIghT, as presented in our paper, can be used to derive spatial graph-based interpretable features from H&E slides and is available as a Nextflow pipeline.
The pipeline comprises the following modules:
- Extracting histopathological features
- Deconvolution of bulkRNAseq data
- Building a multi-task cell type model to predict cell type abundances on a tile-level
- Predicting tile-level cell type abundances using the multi-task models
- Compute spatial features using the tile-level cell type abundances
The training of the cell type models have been perfomed using fresh frozen (FF) slides for the TCGA-SKCM dataset (melanoma) as described in the paper. The trained models are provided here.
See also the figures below.
- Docker:
27.2.0 - Apptainer:
1.0.2 - Nextflow:
24.04.4 build 5917
These were the versions used for testing the pipeline.
- Create apptainer/singularity containers from Docker images:
# 1. save docker as tar or tar.gz (compressed)
docker save joank23/spotlight -o spotlight.tar.gz
# 2. build apptainer (.sif) from docker (.tar)
apptainer build spotlight.sif docker-archive:spotlight.tar.gz
# 1. save docker as tar or tar.gz (compressed)
docker save joank23/immunedeconvr -o immunedeconvr.tar.gz
# 2. build apptainer (.sif) from docker (.tar)
apptainer build immunedeconvr.sif docker-archive:immunedeconvr.tar.gz
- Download retrained models to extract the histopathological features, available from Fu et al., Nat Cancer, 2020 (Retrained_Inception_v4). Once you unzip the folder, extract the files to the
data/checkpoint/Retrained_Inception_v4/folder.
For more information, please read docs/README.md
If you use SPoTLIghT, please cite our paper:
Lapuente-Santana, Ó., Kant, J. & Eduati, F. Integrating histopathology and transcriptomics for spatial tumor microenvironment profiling in a melanoma case study. npj Precis. Onc. 8, 254 (2024). https://doi.org/10.1038/s41698-024-00749-w
