[GM only] Modeling results

[lcolladotor]

After #34, let's generate the following modeling results using the same modeling strategy from the Maynard et al, 2021 paper. That is:

• enrichment (you did this already for the spatial registration) from https://github.com/LieberInstitute/HumanPilot/blob/7049cd42925e00b187c0866f93409196dbcdd526/Analysis/Layer_Guesses/layer_specificity.R#L1424-L1443
• pairwise from https://github.com/LieberInstitute/HumanPilot/blob/7049cd42925e00b187c0866f93409196dbcdd526/Analysis/Layer_Guesses/layer_specificity.R#L1355-L1383
• ANOVA from https://github.com/LieberInstitute/HumanPilot/blob/7049cd42925e00b187c0866f93409196dbcdd526/Analysis/Layer_Guesses/layer_specificity_fstats.R#L24-L85

The enrichment statistics here will be slightly different because for spatial registration you did it across the BayesSpace clusters, while here we'll do it across the pathology labels (only the GM-only spots from case samples).

This could be a great case for using the new function(s) that you are working on with Abby. Because we need the enrichment results right now, we could generate them without the new function(s) and later update them. The next priority would be the pairwise and ANOVA would be the last one.