plot_umap error and problem with mutate() column metrics

[hwenxi]

Hi, Thank you in advance for your help and for providing this great tool. While trying to plot the AUC over the UMAP, I ran into the following error.

Error in `$<-.data.frame`(`*tmp*`, "auc", value = numeric(0)) : 
replacement has 0 rows, data has 48804

I am using a seurat object that has a umap dimensional reduction. After skimming the plot_umap.R file I cannot pinpoint where this error is occuring.

Furthermore, I am not sure if this is relevant or a separate error but I am getting 50+ occurrences of this error when I run the calculate auc. But I am still presented with AUC scores for each of my cell types at the end.

Problem with `mutate()` column `metrics`.
i `metrics = pred %>% map(metric_fun)`.
i While computing binary `precision()`, no predicted events were detected (i.e. `true_positive + false_positive = 0`). 
Precision is undefined in this case, and `NA` will be returned.
Note that 6 true event(s) actually occured for the problematic event level, 'hibd 24h'

Please let me know if you could point me in the right direction for fixing these issues! Thanks.

[jordansquair]

Having trouble reproducing this. Can you provide a bit more info on what you are passing into plot_umap? Are you passing in the entire augur results (i.e., including the y, cell_types, results, etc slots) or just the AUC data frame?

To help reproduce the error. Can you download the Kang2018.rds from here: https://zenodo.org/record/4772064#.YOtN7BNKh24

I've just done the following and I am not getting any issues:

library(Seurat)
library(tidyverse)
library(Augur)

# load in the Kang dataset
sc = readRDS("data/rnaseq/seurat/Kang2018.rds") %>%
  NormalizeData() %>%
  FindVariableFeatures() %>%
  ScaleData() %>%
  RunPCA() %>%
  RunUMAP(dims = 1:10)

# run augur
augur = calculate_auc(sc)

# plot
plot_umap(augur, sc)

Thanks for the note about the warning. This is just for the calculation of precision so not to worry, we will suppress it.

[hwenxi]

Hi, I discovered my error was naming the cell annotations column differently and not specifying the new column name in plot_umap. I am now running into a ggplot error:

> plot_umap(tfaug, twofour, cell_type_col = "reclustered")
Error: Breaks and labels are different lengths
Run `rlang::last_error()` to see where the error occurred.
In addition: Warning message:
`guides(<scale> = FALSE)` is deprecated. Please use `guides(<scale> = "none")` instead. 

Do you have any idea what could be causing this? Thanks for all the help!

[jordansquair]

I am still not quite able to reproduce your error. Are you sure you do not have some NAs in your cell type vector?

[hwenxi]

I'm not sure how to explain this, but I just reran everything exactly the same as I have before and it worked? I am pretty sure I did not change anything and I have not reset my environment since first posting this problem. Hopefully this bug doesn't appear again for me or anyone else...

Thanks Jordan for all the help!