andreasbueckle
commented
3 months ago From @katyb : Yes, use John’s annotations and segmentations for HRApop paper.
A future HRA Workspace could compare both. I added this to https://docs.google.com/document/d/1trF71s4t10TVw6HA1xHPSln_5x7ck-SsuuNrfnNwH48/edit#bookmark=id.4bstbds06yhv k
@katyb @bherr2
I am adding this GH issue so we can track this feature request for the future.
From Yash: Hey Andi, there are two different sources for protein biomarker expression values for sc-proteomics datasets. Data Portal: On the Data Portal, imaging sc-proteomics datasets that have been processed via x+SPRM will have SPRM output files as part of the processed dataset. One of the files labeled "reg001_expr.ome.tiff-cell_channel_meanAll.csv" will have the mean intensity for all biomarkers for whole cell. For now, all CODEX Hickey Intestine datasets that have been processed via [Cytokit + SPRM] pipelines have these files available. I do not recommend using these values for now since there will be a discrepancy between the cell type counts in the data I provided on GitHub vs the cell type counts on the portal. The reason for this is different processing pipelines (cytokit cell segmentation on portal vs different cell segmentation approach by TMC). Hope is that once the portal has the cell types available, all the information can be directly derived from the portal itself. Data Provider: For the two organ sets I provided: Intestines and skin, the biomarker expressions can be taken from the data they provided. For the Intestine Hickey datasets, that data is available as part of their Dryad dataset (https://datadryad.org/stash/dataset/doi:10.5061/dryad.pk0p2ngrf). For skin datasets, I don't think we have that information. For intestine data, I recommend using the expression values from the dryad dataset since that is where we get the cell type counts as well.