search

EDePasquale / DoubletDecon

A tool for removing doublets from single-cell RNA-seq data
69 stars 19 forks source link

"Unable to perform mcl function for blacklist clustering" while using GUI #28

Open carlyrosewilling opened 4 years ago

carlyrosewilling commented 4 years ago

Hi there, I saw a similar issue opened when running the package command line, but I am running the GUI. In the console showing the current progress, I get the following warning/error sequence:

Creating original data heatmap... Combining similar clusters... Error in mcl(blacklist, addLoops = FALSE)$Cluster : $ operator is invalid for atomic vectors [1] "Unable to perform mcl function for blacklist clustering, please try a different rhop." Warning: Error in Blacklist_Groups: 83: stop 82: Blacklist_Groups [Blacklist_Groups.R#63] 75: Main_Doublet_Decon_UI [Main_Doublet_Decon_UI.R#89] 74: observeEventHandler [/private/var/folders/x8/frw_g_ld5hg9m4l73c6gx0vc0000gn/T/RtmpO73P82/shinyapp365713293057/a81cdc2aea5742c08e5fc3fa66d47698-master/app.R#759] 3: runApp 2: runUrl 1: shiny::runGist

After this, I cannot access the cluster similarity viewer, results summary, or interactive heatmap. Just looking for some guidance as to how to approach this and be able to visualize the results/recluster after doublet removal.

Thank you in advance!

EDePasquale commented 4 years ago

Hi,

This error most commonly occurs when the value you selected for the cluster merging parameter ρ' (‘rhop’ in the parameter list) causes too much or too little merging of the clusters, leading to an error in the Markov clustering portion of the algorithm. Altering the value of this parameter according to the following instructions should solve the issue:

  1. Begin by starting the DoubletDecon desktop application and uploading all relevant data files. Keep the remaining parameters at their default values.
  2. Select the “Test for rho-prime values” button at the bottom of the initial screen. A new tab will automatically open when processing is completed, which will display a number of cluster merging heatmaps along with the associated ρ' values used to generate those heatmaps. (Note: Please be patient, as the processing for this step may take several minutes.)
  3. Examine the clustering merging heatmap that was automatically generated in conjunction with previously identified cluster labels to assess the appropriateness of merging. When merging clusters, the goal is to combine highly similar clusters to improve the formation of reference clusters for deconvolution and to generate transcriptionally distinct synthetic doublets, both of which improve the overall accuracy of the method. However, by merging two clusters the software will no longer be able to detect doublets formed between those two clusters. This is why a thorough understanding of the cell cluster identifications and relevant biology are critical to using DoubletDecon, as it may be very important to remove doublets of some classes while retention of other doublet types may not affect downstream analyses. Please note this value for future steps.
  4. Close and restart the DoubletDecon application before moving on to the following steps.

The DoubletDecon UI has a built-in function for displaying all possible values of ρ' to make it easier to choose one within in the range. Once selected, make sure to evaluate the appropriateness of the clustering and adjust up or down depending on whether you want less or more cluster merging, respectively.

Please let me know if this doesn't solve the issue!

Best, Erica