This repository contains code for Visium data processing, downstream analysis, and figure generation for the study titled "Profiling of epithelial functional states and fibroblast phenotypes in hormone therapy-naive localized prostate cancer."
The Visium data was processed using the scanpy package (v1.9.1) and spatial deconvolution was performed using Cell2Location (v0.1.3).
- In-house Visium samples
- Batch 1: 2 samples - Visium FFPE V1 (2 x 6.5x6.5mm reactions)
- Batch 2: 8 samples - Visium FFPE V1 (8 x 6.5x6.5mm reactions)
- Batch 3: 4 samples - Visium FFPE V2 (2 x 11x11mm reactions)
- Published data from Joakim's lab (PMID 35948798)
- Contains parameter files for each script, sample sheet & other ad hoc files
- Contains images, anndata objects (filtered, log_norm, raw, raw_beature_bc_matrix), per-spot histopathology annotation (only for samples in the paper)
- Contains EDF and ESF files for pca_visium
- Contains gene lists (e.g., Cepo derived signatures for pnCAFs and Glial cells), and published data (i.e. PMID 35948798)
- Script outputs are saved to corresponding results dir
- Includes results related to:
- Basic processing and QC (found in basic_qc/, clustering/, and marker_genes/)
- Cell2Location analysis (found in 02_cell2location/) - the contents of the subdirectories should be self-explanatory.
- NOTA BENE - most analyses related to the PCa manuscript can be found here!
- Cell2Location colocalisation analyses (inhouse and published PMID 35948798 - not relevant to PCa manuscript), H&E plots, and pnCAF and Glial scanpy scores (found in 04_colocalisation_inhouse_and_published)
- Includes results related to:
01_basic_qc/
- Initial pre-processing (without filtering spots), QC & visualisation
02_cell2location/
- 01_h5ad_conversion.R:
- Convert Seurat object to .h5ad
- 02_train_model.py:
- Train cell2location model on PCa scRNA-seq reference (cell type minor)
- 03a_cell_type_mapping.py:
- Map cell types (cell type minor)
- 04_visualize_results.py:
- Visualize results (cell type minor)
- 05_train_model_major.py:
- Train cell2location model on PCa scRNA-seq reference (cell type major)
- 06a_cell_type_mapping_major.py:
- Map cell types (cell type major)
- 07_visualize_results_major.py:
- Visualize results (cell type major)
- 08_visualise_proportions_minor.R
- 09_visualise_proportions_major.R
- 10_colocalisation_major.R:
- Summarise cell2location results, & calculate spot-level cell-cell correlations (grouped by sample id & histological feature; cell type major)
- 11_colocalisation_minor.R:
- Summarise cell2location results, & calculate spot-level cell-cell correlations (grouped by sample id & histological feature; cell type minor)
- 12_train_model_minor_mal.py:
- Train cell2location model on PCa scRNA-seq reference (cell type minor mal - i.e. a combination of cell type minor annotation and malignant spectrum annotation for Epithelial cells)
- 13a_cell_type_mapping_minor_mal.py:
- Map cell types (cell type minor mal)
- 14_visualize_results_minor_mal.py:
- Visualize results (cell type minor mal)
- 15_colocalisation_minor_mal.R:
- Summarise cell2location results, & calculate spot-level cell-cell correlations (grouped by sample id & histological feature; cell type minor mal)
- 16_train_model_major_temp.py:
- Train cell2location model on PCa scRNA-seq reference (cell type major temp - i.e. use minor annotation for CAFs and major annotation for all other cell types)
- 17a_cell_type_mapping_major_temp.py:
- Map cell types (cell type major temp)
- 18_visualize_results_major_temp.py:
- Visualize results (cell type major temp)
- 19_colocalisation_major_temp.R:
- Summarise cell2location results, & calculate spot-level cell-cell correlations (grouped by sample id & histological feature; cell type major temp)
- Processed Visium V1 sequencing and image data is available for in-browser exploration and download through the CELLxGENE portal.
- Raw Visium data will be deposited in the European Genome-Phenome Archive (EGA).
For further enquires, please either raise an issue via GitHub or email John Reeves (Data Manager - j.reeves(at)garvan.org.au) or Alexander Swarbrick (Lab Head - a.swarbrick(at)garvan.org.au).
- Further deidentify Cansto IDs - need to be removed from scripts and .dvc files!
- Create a new repo after updates have been made & remove scripts/data not relevant to paper (i.e. analysis of data from PMID 35948798)
- Add CELLxGENE links for V1 (when available)
- Add info on where Visium V2 data can be found
- Add EGA accession number (when available)
- Reference Visium data DVC GitHub repo
- Reference scRNA-seq GitHub repo