"undefined columns error"

[egarren]

Running into this error after the analyze(obj) step and not sure what's the cause. For background, I have imported a Seurat object and am using signatures from msigdb. Any idea what could be causing the error? Thank you in advance!

vision.obj <- analyze(vision.obj)
Beginning Analysis

Performing micro-pooling on 28703 cells with a target pool size of 5
Using supplied latent space with 2 components
Performing initial coarse-clustering...
Further partitioning coarse clusters...
Micro-pooling completed reducing 28703 cells into 5739 pools
Aggregating data using assigned pools...completed

Clustering cells...completed

Projecting data into 2 dimensions...
Running method 1/1: tSNE30 ...

Evaluating signature scores on cells...

|============================================================================================================| 100%, Elapsed 00:14
Evaluating signature-gene importance...

|============================================================================================================| 100%, Elapsed 00:54
Creating 5 background signature groups with the following parameters:
sigSize sigBalance
1 9 0.9902046
2 24 0.9742071
3 70 0.9571728
4 254 0.9888702
5 276 0.5600931
signatures per group: 3000
Computing KNN Cell Graph in the Latent Space...

Evaluating local consistency of signatures in latent space...

|============================================================================================================| 100%, Elapsed 00:51
|============================================================================================================| 100%, Elapsed 01:24
|============================================================================================================| 100%, Elapsed 50:51
|============================================================================================================| 100%, Elapsed 00:01
Clustering signatures...

Error in [.data.frame(sigMatrix, , keep_significant, drop = FALSE) :
undefined columns selected

[mattjones315]

Hello,

We've seen errors like this when the signatures passed in have duplicate names. To help us diagnose this issue, can you:

1. Verify that each signature has a unique name? You can check this by looking at the Signatures stored in the @sigdata slot. Look at the @name slot in each Signature object in the @sigdata slot and verify that each name appears once and only once.
2. Provide us with a minimal reproducible example that we can run.

Thank you!

-Matt

[egarren]

Thanks Matt! I checked and my signature names appear to be distinct. I ran the following to check:

sum(duplicated(names(vision.obj@sigData)))
[1] 0

Here are some snipets of my code:

library(msigdb)
signatures<-msigdb.genesets(type="symbols", species="mouse", return.data.frame=FALSE)
write.gmt(signatures,"mouse.gmt")
count.data <- GetAssayData(object = pbmc[["SCT"]], slot = "counts")
new.seurat.object <- SetAssayData(object = pbmc,slot = "scale.data",
new.data = as.matrix(x = count.data),assay = "RNA")
vision.obj <- Vision(new.seurat.object,signatures ="mouse.gmt",pool = T,
cellsPerPartition = 5, dimRed<-"umap")
vision.obj <- analyze(vision.obj)

[dpcook]

Have come across the same issue and tracked down the specific signature causing the issue. I'm iterating through subsets of a complete dataset, running vision on each subset. Made it through 8 subsets, but on the 9th, I'm getting:

vision <- Vision(seurat[[i]],
                   signatures = signatures,
                   sig_gene_threshold = 0.025,
                   projection_methods=NULL)
vision <- analyze(vision)

Clustering signatures...

Error in `[.data.frame`(sigMatrix, , keep_significant, drop = FALSE) : 
  undefined columns selected`

Can't track down the source of the error on this one specific data set. It's a single signature, so duplicated names shouldn't be an issue. I'll keep looking around, but here's access to the data that should reproduce the error.

Signature: https://drive.google.com/file/d/1wXD3KzUw4-KwyGlu0YEVnQWSU3HJVzDH/view?usp=sharing
Seurat object (~945 cells): https://drive.google.com/file/d/1HiTy6yRgedpOQJQf11PV123RlnnXE0xs/view?usp=sharing

[dpcook]

Update: Nevermind, turns out there are no duplicated names in the signatures themselves, but I had a seurat@meta.data column with the identical name to the signature ("Hypoxia"). Removing the duplicated name fixed it.