Downstream Analysis of Data from Spatial Transcriptomics Profiling of Peritoneal Metastases Using BGI Stereo-seq

This study applied spatial transcriptomic profiling to tissue samples from two patients (`NCCMC393` - CRC, `NCCMC400` - struma ovarii) with peritoneal metastases using the BGI Spatial Enhanced Resolution Omics-sequencing (Stereo-seq) platform. The dataset provides spatially resolved gene expression maps for exploring the molecular architecture of peritoneal metastases. Fresh-frozen tissue sections were H&E stained, imaged, and processed for spatially resolved transcriptome profiling. Libraries were prepared and sequenced to generate High-resolution gene expression maps (`results/0.saw/NCCMC393_NCCMC400.tissue.gef`) for downstream analysis. Folder structure:A) `ref_datasets.zip`: Reference single cell RNA-seq (scRNA-seq) for cell type deconvolution using `cell2location` 1) `CRC/`: We subset adolescent and adult colorectal cells from gut cell atlas dataset [1] 2) `Thyroid/`: NCBI GEO dataset `GSE193581` - cells from thryoid tissues of healthy people, Anaplastic Thyroid Carcinoma (ATC) and Papillary Thyroid Carcinoma (PTC) patients. See Lu L et al. (2023) [2]B) `code.zip` and `results.zip`: jupyter notebook file, and respective output files for analyses steps: 0) `0.saw` - `NCCMC393_NCCMC400.tif` - tif file for ssDNA staining from Stereo-seq tissue section. Sections NCCMC393_pre and NCCMC400_pre have an overlapping region. - `NCCMC393_NCCMC400.tissue.gef` - Gene expression map 1) `1.export_square_bin_info` - Export AnnData file for BIN100s, from gene expression map 2) `2.split_BIN100_scanpy` - Use a straight line to split 3) `3.prepare_scRNAseq_ref` - Filter cells from scRNA-seq reference datasets, and unify names of cell types 4) `4.train_cell2location_ref_model` - Using cleaned scRNA-seq reference datasets to train `` model and estimate gene expression signature of cell types in reference datasets 5) `5a.fine_split_of_NCCMC393` - Use cellcharter to do spatial clustering, and remove overlapping regions between two clusters. 6) `6a.cell_type_deconvolution_NCCMC393` - cell type deconvolution by cell2location, using thyroid reference dataset `6b.cell_type_deconvolution_NCCMC400` - cell type deconvolution by cell2location, using CRC reference dataset 7) `7.merge_cell_type_deconvolution_results` - Merge both results in 6) into a single AnnData object 8) `8.SERPINE1_heatmap` - Generate hotspot for: (1) High fibroblast proportion (2) High SERPINE1 expression (3) merged References:[1] Oliver, A.J., Huang, N., Bartolome-Casado, R., Li, R., Koplev, S., Nilsen, H.R., Moy, M., Cakir, B., Polanski, K., Gudiño, V., et al. (2024). Single-cell integration reveals metaplasia in inflammatory gut diseases. Nature 635, 699–707. https://doi.org/10.1038/s41586-024-07571-1.[2] Lu, L., Wang, J.R., Henderson, Y.C., Bai, S., Yang, J., Hu, M., Shiau, C.-K., Pan, T., Yan, Y., Tran, T.M., et al. (2023). Anaplastic transformation in thyroid cancer revealed by single-cell transcriptomics. J Clin Invest 133, e169653. https://doi.org/10.1172/JCI169653.