Semi-supervised clustering on this dataset

[abspangler13]

Use marker genes from pilot dataset to call layers in this new dataset

[lmweber]

Hi @abspangler13 , posting some links here as we discussed just now.

My code from our previous DLPFC paper is in these .Rmd files in the HumanPilot repo:

• https://github.com/LieberInstitute/HumanPilot/blob/master/Analysis/SpatialDE_clustering.Rmd
• https://github.com/LieberInstitute/HumanPilot/blob/master/Analysis/SpatialDE_genes_analysis.Rmd
• https://github.com/LieberInstitute/HumanPilot/blob/master/Analysis/SpatialDE_statistics.Rmd
• https://github.com/LieberInstitute/HumanPilot/blob/master/Analysis/SpatialDE_subsampling_comparison.Rmd

I think the semi-supervised clustering is all in the first one, so:

• https://github.com/LieberInstitute/HumanPilot/blob/master/Analysis/SpatialDE_clustering.Rmd

There are several subsections, which run the different types of clustering and calculate the evaluation metrics.

You can find the semi-supervised clustering parts by searching for "semi" and "pseudobulk".

I organized the clustering code into functions to make it easier to run all the different scenarios at that time, but if you are running a single type of clustering now (semi-supervised) then it may be easier to just copy out the parts of the function that you need, instead of re-using the whole function.

[lcolladotor]

Wait to do this until #41 is done.