Speed ups

[will-NYGC]

Hi,

I have a more complex experimental design with several time-points and assay types and I'd like to give this method a try. Because of the design, I have 100,00's of 200bp genomic windows annotation with 1 of 26 classes (>= 200 windows for each). Obviously, trying to reduce the classes and number of windows could help but are there any other speed ups you'd suggest? How much performance loss is there if I do say 200 ADMM iterations instead?

best.