plotOntologyHeatmapCollapsed function got an error

[Angel-Wei]

Hi, I ran into an error when I called cao$plotOntologyHeatmapCollapsed function. I got an error

Error in `chr_as_locations()`:
! Can't rename columns that don't exist.
✖ Column `qvalues` doesn't exist.

Before I come to this step, everything ran well and functions cao$estimateDEPerCellType and cao$estimateOntology (I used 'GSEA')also worked. Here's some output info I got from rlang::last_error():

> pdf(paste0(rep.name, '/', rep.name,"_plotOntologyHeatmapCollapsed_up_in_MUT.pdf"), height=15, width=10)
> print(cao$plotOntologyHeatmapCollapsed(
+   name="GSEA", genes="up", n=50, clust.method="ward.D", size.range=c(1, 4)) + ggtitle("Up in MUT"))
Error in (function (cond)  : 
  error in evaluating the argument 'x' in selecting a method for function 'print': Can't rename columns that don't exist.
✖ Column `qvalues` doesn't exist.
> rlang::last_error()
<error/vctrs_error_subscript_oob>
Error in `chr_as_locations()`:
! Can't rename columns that don't exist.
✖ Column `qvalues` doesn't exist.
---
Backtrace:
  1. cao$plotOntologyHeatmapCollapsed(...)
  6. dplyr:::rename.data.frame(., geneID = core_enrichment, qvalue = qvalues)
  7. tidyselect::eval_rename(expr(c(...)), .data)
  8. tidyselect:::rename_impl(...)
  9. tidyselect:::eval_select_impl(...)
 18. tidyselect:::vars_select_eval(...)
 19. tidyselect:::walk_data_tree(expr, data_mask, context_mask, error_call)
 20. tidyselect:::eval_c(expr, data_mask, context_mask)
 21. tidyselect:::reduce_sels(node, data_mask, context_mask, init = init)
 22. tidyselect:::walk_data_tree(new, data_mask, context_mask)
 23. tidyselect:::as_indices_sel_impl(...)
 24. tidyselect:::as_indices_impl(x, vars, call = call, strict = strict)
 25. tidyselect:::chr_as_locations(x, vars, call = call)
Run `rlang::last_trace()` to see the full context.

I wonder is there anything I can do to fix this problem to generate GSEA results? Thank you so much for the help in advance!

Originally posted by @Angel-Wei in #21 (comment)

[rrydbirk]

Hi @Angel-Wei

Thanks for your interest in Cacoa. We're aware of this error, it should be fixed in the dev version. You can install this using devtools::install_github("kharchenkolab/cacoa", ref="dev"). If you saved your Cacoa object, you need to update it after installing, e.g., cao %<>% Cacoa$new().

Could you let me know if this solves your problem?

BR

Rasmus

[Angel-Wei]

Hi Rasmus,

Sorry for the late reply. I actually tried to install the dev version. However, it gave me error and warning messages like this:

I wonder is there any way to address this issue?

[rrydbirk]

Hi @Angel-Wei
We've made some changes to sccore, as stated in the error message v. 1.0.3 is required. Could you do a manual install.packages("sccore") followed by cacoa dev install? That should do the trick.

BR

Rasmus

[Angel-Wei]

Oh hi Rasmus! I think the install.packages("sccore") did not work in my case, which just installed the same version as I had. However, I found that the latest version can be installed via devtools:

Now my problem is solved and I can install the latest version of cacoa and plot the heatmap of GSEA. Thank you so much for the help!

[rrydbirk]

Hi @Angel-Wei
Thanks for the feedback, glad it worked for you!

@evanbiederstedt, am I missing something here, why is sccore v. 1.0.3 not installing per default?

/Rasmus

[Angel-Wei]

Hi Rasmus,

Just an update of this function error. I realized that it worked after I re-install the latest Cacoa and use
cao %<>% Cacoa$new() (R 4.2.0 and cacoa 0.4.0) on a previously saved old Cacoa object (created and saved under R 4.1.3 and cacoa 0.4.0).
However, if I re-run Cacoa pipeline using the latest version (R 4.2.0 and cacoa 0.4.0), for the new cao object, I still encountered errors like this:

> traceback()
17: stop(fallback)
16: signal_abort(cnd)
15: cnd_signal(cnd)
14: value[[3L]](cond)
13: tryCatchOne(expr, names, parentenv, handlers[[1L]])
12: tryCatchList(expr, classes, parentenv, handlers)
11: tryCatch(with_entraced_errors(expr), vctrs_error_subscript = function(cnd) {
        cnd$subscript_action <- subscript_action(type)
        cnd$subscript_elt <- "column"
        cnd_signal(cnd)
    })
10: with_subscript_errors(vars_select_eval(vars, expr, strict = strict, 
        data = x, name_spec = name_spec, uniquely_named = uniquely_named, 
        allow_rename = allow_rename, type = type, error_call = error_call), 
        type = type)
9: eval_select_impl(x, names, {
       {
           sel
       }
   }, strict = strict, name_spec = name_spec, type = "rename", error_call = error_call)
8: rename_impl(data, names(data), as_quosure(expr, env), strict = strict, 
       name_spec = name_spec, error_call = error_call)
7: tidyselect::eval_rename(expr(c(...)), .data)
6: rename.data.frame(., geneID = core_enrichment, qvalue = qvalues)
5: rename(., geneID = core_enrichment, qvalue = qvalues)
4: ont.res %<>% rename(geneID = core_enrichment, qvalue = qvalues)
3: private$getOntologyPvalueResults(name = name, genes = genes, 
       subtype = subtype, cell.subgroups = cell.subgroups, p.adj = p.adj, 
       q.value = q.value, min.genes = min.genes, readjust.p = readjust.p, 
       p.adjust.method = p.adjust.method)
2: private$getOntologyHeatmapInfo(name = name, genes = genes, subtype = subtype, 
       p.adj = p.adj, q.value = q.value, min.genes = min.genes, 
       selection = selection, cluster = cluster, cell.subgroups = cell.subgroups, 
       only.family.children = only.family.children, clust.naming = clust.naming, 
       description.regex = description.regex, description.exclude.regex = description.exclude.regex, 
       readjust.p = readjust.p, p.adjust.method = p.adjust.method)
1: cao$plotOntologyHeatmapCollapsed(name = "GSEA", genes = "up", 
       n = 50, clust.method = "ward.D", size.range = c(1, 4))

Not sure about reasons and it could still work in the first case. But like to see any feedback if there is any! Thank you so much for the information provided!

[rrydbirk]

Hi @Angel-Wei

This was fixed some time ago here.

Could you check cao$.__enclos_env__$private$getOntologyPvalueResults and see if it's correct there in the function definition? It should be

    if (type == "GSEA") {
        if ((nrow(ont.res) == 0) && !("core_enrichment" %in% 
            colnames(ont.res))) {
            ont.res$core_enrichment <- character()
        }
        ont.res %<>% rename(geneID = core_enrichment)

I just checked our copy constructor, private functions should be updated as well with cao %<>% Cacoa$new().

[Angel-Wei]

Hi Rasmus, so I checked cao$.__enclos_env__$private$getOntologyPvalueResults in case 2 I mentioned above (R 4.2.0, Cacoa 0.4.0). It shows ont.res %<>% rename(geneID = core_enrichment, qvalue = qvalues), which is the line 3022 deleted in pull request #42. It stayed the same even if I update private functions with cao %<>% Cacoa$new()
Not sure about the reasons, but I guess I'll stick to the first case I mentioned above to use the function. Thank you so much for your patience in this!

[rrydbirk]

No worries, as long as it's working, all is good :-) Thanks for using our software!