To Do List

[alexanderlewis99]

(1) repr_metacells: return_as_anndata option, return_as_df option (2) Function that computes a gExpr signature by normalizing the size of every cluster. Uses rep_subsample for each cluster to shrink each cluster down to the size of the smallest given cluster. (3) Plot OncoMatch function • Give obs column names as a list corresponding to each AnnData • Specify cutoff point (4) plot.msviper translated into python (5) Make rank_norm multicore

[alexanderlewis99]

(i) Make it clear spearman = closeness (ii) Option to combine p-values (e.g. between spearman and Stouffer) by using a general signature function that incorporates multiple methods with fischer p-value (iii) Option to rank by neg-log p-values (break ties using score) instead of just by score (iv) Option to find_top_mrs using p-value cutoff (e.g. 0.05) or using % (e.g. top 5%) of total MRs instead of by top N (v) Add progress bar descriptions (e.g. "Stouffer")

[alexanderlewis99]

For reprmetacells: (a) Add ability to specify multiple sizes for different clusters, e.g. [100, 100, 20] for clusters of sizes [1000, 500, 90]. Also allow users to do that by proportion by giving values less than 1, e.g. [0.25, 0.25, 0.25] to do 25% of cells per cluster. (b) Add a parameter called prefix, which is the name of the samples in the matrix. Should be cluster name when cluster labels given. Currently it's "metacell" and then 0 to # of metacells.