Questions about matched gene sets

[HelloWorldLTY]

Hi, thanks for your great work. I wonder if it is possible to share more details of selecting matched control genes. e.g., how the threshold of matched mean/variance/other factors affects downstream analysis. Are they hyper-parameters that we can tune? Thanks.

[martinjzhang]

We match by 400 (20*20) mean-variance bins.

• Here is where the mean-variance bins are computed: https://github.com/martinjzhang/scDRS/blob/c4dfcc09b8bfda7c26e5bb8674560c5bf9048bab/bin/scdrs#L248-L250
• Details of scdrs.preprocess here https://martinjzhang.github.io/scDRS/reference/scdrs.preprocess.html
• It will generate a column of mean_var in adata.uns[“SCDRS_PARAM”]["GENE_STATS"], and will be later used in scdrs.score_cell: https://github.com/martinjzhang/scDRS/blob/c4dfcc09b8bfda7c26e5bb8674560c5bf9048bab/bin/scdrs#L264-L274

[HelloWorldLTY]

Got it, thanks a lot!