Updated metaviper/gsea enrichment

[allaway]

By drug, this time. This is more interesting than the previous one, but based on the network plots it's pretty clear that this analysis effectively disregards any drugs with only a few listed targets. I included all drugs, though, but pretty much all of the highest-enriched drugs are have a broad range of protein targets

[sgosline]

Only change is to move to results directory. Pretty exciting stuff!

[allaway]

Whoops, didn't realize it was in the wrong directory. just fixed that!

[sgosline]

Cool, that was my only concern. simpler is better :)

[sgosline]

Re-cap of offline conversation: I think it'd be helpful to include bargraph of number of targets for each drug alongside the enrichment plots

Additional question - is the foldchange in the network actually correlation? If so this is interesting.

[allaway]

the latest version of this pr has this code: ``` tab <- lapply(names(res_complt), function(x){ foo <- res_complt[[x]]@Result %>% mutate(latent_var = x) %>% select(latent_var, everything()) }) %>% bind_rows() # lv_drug_tab <- synBuildTable("Latent Variable Drug Set Enrichment Analysis", "syn21046734", tab) # synStore(lv_drug_tab) DT::datatable(tab) ``` pushed to this table: https://www.synapse.org/#!Synapse:syn21442304/tables/

…

On Mon, Dec 16, 2019 at 2:18 PM Sara JC Gosline ***@***.***> wrote: ***@***.**** requested changes on this pull request. ------------------------------ In 30-viper-drug-enrichment-bydrug.Rmd <#92 (comment)> : > +}, mc.cores = parallel::detectCores()) + +names(res) <- tidy$latent_var + +##determine if no results for a given LV +empty_idx <- lapply(res, function(x){ + ***@***.***)==0){ + TRUE + }else{ + FALSE + } +}) + +##remove if no results +res_complt <- res[empty_idx==F] + Can you post a table of the the drug x LV enrichment scores and p-values to Synapse? I'd like to pull it down in the correlation analysis. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#92?email_source=notifications&email_token=AE3WNSCETD3KIO5VIVMYXDDQY75B5A5CNFSM4J26LWW2YY3PNVWWK3TUL52HS4DFWFIHK3DMKJSXC5LFON2FEZLWNFSXPKTDN5WW2ZLOORPWSZGOCPL3VZI#pullrequestreview-332905189>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AE3WNSFOYF62NQYAPPLARZTQY75B5ANCNFSM4J26LWWQ> .

[allaway]

Re-cap of offline conversation: I think it'd be helpful to include bargraph of number of targets for each drug alongside the enrichment plots

Additional question - is the foldchange in the network actually correlation? If so this is interesting.

@sgosline barplots are added in the latest version of this branch!

[sgosline]

looks good, thanks for the updates!