Combining sequencing platforms - 10x and Cel-seq2 - Error in if (trim > 0 && n) { : missing value where TRUE/FALSE needed

[reberya]

Hello,

First off, thanks for putting together an awesome tool. I'm a new user to rliger and was wondering if you had any guidance on how to handle integrating datasets from different analysis platforms - specifically I'm working with 10x data and CEL-Seq2 data. What I've tried thus far is to merge these objects into a single Seurat object and proceed with the integration strategy as outlined here.

toMerge = a list of Seurat objects from 10x data
kild_cs = a Seurat object from cel-seq2 data

# combine the seurat objects
MERGED <- merge(toMerge[[1]], 
              y = c(toMerge[[2]],toMerge[[5]],toMerge[[8]],toMerge[[11]],toMerge[[14]],
                    toMerge[[3]],toMerge[[6]],toMerge[[9]],toMerge[[12]],toMerge[[15]],
                    toMerge[[4]],toMerge[[7]],toMerge[[10]],toMerge[[13]],toMerge[[16]],
                    kild_cs))

# process with rliger
MERGED %<>% NormalizeData(.) %>%
  FindVariableFeatures(.) %>%
  ScaleData(., split.by = "SAMPLE", do.center = FALSE) %>% 
  RunOptimizeALS(., k = 20, lambda = 5, split.by = "SAMPLE") %>%
  RunQuantileNorm(., split.by = "SAMPLE") %>%
  FindNeighbors(., reduction = "iNMF", dims = 1:20)

However, this kicks out the error shown during the RunOptimizeALS() call:

Performing log-normalization
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Calculating gene variances
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Calculating feature variances of standardized and clipped values
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Scaling data matrix
Scaling data from split dong_GSM4088774
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4088775
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4088776
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4088777
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4088778
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4088779
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4088780
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4088781
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4088782
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4088783
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4654669
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4654670
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4654671
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4654672
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4654673
  |=================================================================================================================================================================================| 100%
Scaling data from split dong_GSM4654674
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_000CGH
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_000FQM
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_000GGU
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_NB060
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_NB086
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_NB098
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_NB106
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_NB107
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_NB123
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_NB124
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_NB125
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_NB130
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_NB132
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_NB138
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_NB151
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_NB152
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_PD42184
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_PD42752-1
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_PD42752-2
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_PD43255
  |=================================================================================================================================================================================| 100%
Scaling data from split kild_PD46693
  |=================================================================================================================================================================================| 100%
  |                                                                                                                                                                                 |   0%
Error in if (trim > 0 && n) { : missing value where TRUE/FALSE needed

I'm fairly certain this is occuring due to discrepancies between the 10x data and the cel-seq2 data since RunOptimizeALS() will finish if I remove the cel-seq2 seurat object. I also can successfully run the ifnb example provided. Happy to provide a reproducible example or share the dataset I'm trying to integrate. Any thoughts would be appreciated. THanks!

[skpalan]

Hi, thanks for your comments! Can you check if the 10x and cel-seq2 data share any genes or features? Also please check if the number of the variable genes selected by FindVariableFeatures is reasonable.

[reberya]

Thanks for the fast reply!

They do share genes.

> head(rownames(toMerge[[1]])) 
[1] "TSPAN6"   "DPM1"     "SCYL3"    "C1orf112" "FGR"      "CFH"      

> head(rownames(kild_cs)) 
[1] "TSPAN6"   "TNMD"     "DPM1"     "SCYL3"    "C1orf112" "FGR"  
    
> table(rownames(toMerge[[1]]) %in% rownames(kild_cs))  
FALSE  TRUE     
85 17954

The number of variable genes seems reasonable

> nbl <- merge(toMerge[[1]], 
+               y = c(toMerge[[2]],toMerge[[5]],toMerge[[8]],toMerge[[11]],toMerge[[14]],
+                     toMerge[[3]],toMerge[[6]],toMerge[[9]],toMerge[[12]],toMerge[[15]],
+                     toMerge[[4]],toMerge[[7]],toMerge[[10]],toMerge[[13]],toMerge[[16]],
+                     kild_10x, kild_cs))
Warning in CheckDuplicateCellNames(object.list = objects) :
  Some cell names are duplicated across objects provided. Renaming to enforce unique cell names.
> 
> test <- nbl %>% 
+   NormalizeData(.) %>%
+   FindVariableFeatures(.)
Performing log-normalization
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Calculating gene variances
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Calculating feature variances of standardized and clipped values
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|

> length(test@[email protected])
[1] 2000
> head(test@[email protected])
[1] "HBB"   "HBA2"  "HBA1"  "IGKC"  "IGHG1" "IGHG4"