Remaining problem in unisusie pipeline

[SamuelSun2000]

@hsun3163 Please note that uni_susie_2 in https://github.com/cumc/xqtl-pipeline/blob/main/code/fine_mapping/SuSiE/SuSiE.ipynb cannot be successfully executed yet. There are two problems:

1. Some of the rds files will be blank, and cannot be handled by readRDS() function. Adding improt os; skip_if(os.path.getsize(_input) == 0) may help.
2. Because now output rds is only one susie_object, there will be errors about the variable susie_list in uni_susie_2. We no longer can use susie_list[[i]]$pip to extract pip. Instead, we should directly use susie_list$pip.

Because I currently don't know the strategy for correcting this step -- whether I should make a vcf for per gene per tissue, or integrate the result per gene in all tissues. So I leave this issue here as a reminder.

[gaow]

@SamuelSun2000 could you provide me a minimal working example for this? (remember the style of the RNA-seq protocol you have tried last year we have the MWE). I'm swamped today until late in the PM .. but this one is now on the top of my TODO to finalize so others can use.

[gaow]

I see. @SamuelSun2000 the MWE you provided me can they reproduce these issues?

[SamuelSun2000]

Yes, both of them.

[hsun3163]

I was thinking, perhaps we should take away the uni_susie_2/3 altogether. And make the transform to vcf step another utility in the susie_post_proc notebook. Along with the code that produces the three summary tables.

[gaow]

yes i remember your proposal. let me take a look maybe this evening or tmr