spaTransfer: transfer learning for single-cell and spatial transcriptomics data using non-negative matrix factorization

This repository contains the processed data for the analyses associated with the manuscript "spaTransfer: transfer learning for single-cell and spatial transcriptomics data using non-negative matrix factorization". Once unzipped, the processed-data folder contains the following folders: dlpfc: One folder for each of the dlpfc samples used in the manuscript (6 total), containing: *-raw-SPE.RDS: SpatialExperiment objects containing the pre-QC Xenium raw counts matrices for each sample *-filt-SPE.RDS: SpatialExperiment objects containing the post-QC Xenium counts matrices for each sample *-filt-predLabels-SPE-bayesspace-logcounts.RDS: SpatialExperiment objects containing the post-QC Xenium counts matrices with spaTransfer domain predictions stored in the colData. traam-outputs, a folder containing the spaTransfer outputs for the analyses in Figure 2. 04_methodComparisons: unsupervised_xenium_comparisons: One folder for each of the 5 unsupervised clustering methods benchmarked in Figure 4b. Each folder contains either SpatialExperiment or AnnData objects with the clustering method's labels stored in the colData. *_labels.csv and *_metrics.csv: Labels and internal metrics for each of the clustering methods, respectively. visium_comparisons: One folder for each of the 3 label transfer methods benchmarked in Figure 4a. Eaach folder contains SpatialExperiment objects with the label transfer method's labels stored in the colData. *ARI*.csv: files contain ARI and NMI values for the label transfer methods. hpc: One folder for each of the hpc samples used in the manuscript (6 total), containing: *-raw-SPE.RDS: SpatialExperiment objects containing the pre-QC Xenium raw counts matrices for each sample *-filt-SPE.RDS: SpatialExperiment objects containing the post-QC Xenium counts matrices for each sample *-filt-predLabels-SPE-bayesspace-logcounts.RDS: SpatialExperiment objects containing the post-QC Xenium counts matrices with spaTransfer domain predictions stored in the colData. traam-outputs, a folder containing the spaTransfer outputs for the analyses in Figure 3. custom-panel-pseudobulkedByDomain-SPE.RDS: pseudobulked SpatialExperiment object used Figure S14-S16.