errors during cell classification

[carloelle]

Hi,

I have a Seurat object named seurat_test

An object of class Seurat 
20387 features across 16489 samples within 1 assay 
Active assay: RNA (20387 features, 2000 variable features)
 1 dimensional reduction calculated: pca

and a trained scPred object

✓  Prediction variable = CellType 
✓  Discriminant features per cell type
✓  Training model(s)
Summary

|Cell type                    |     n| Features|Method    |   ROC|  Sens|  Spec|
|:----------------------------|-----:|--------:|:---------|-----:|-----:|-----:|
|CD14+ Monocyte               |  1458|       50|svmRadial | 0.996| 0.760| 0.994|
|CD19+ B                      |  4184|       50|svmRadial | 0.963| 0.674| 0.998|
|CD34+                        |   141|       50|svmRadial | 0.999| 0.893| 1.000|
|CD4+ T Helper2               |    69|       50|svmRadial | 0.693| 0.000| 1.000|
|CD4+/CD25 T Reg              |  4587|       50|svmRadial | 0.911| 0.321| 0.987|
|CD4+/CD45RA+/CD25- Naive T   |  1392|       50|svmRadial | 0.814| 0.002| 1.000|
|CD4+/CD45RO+ Memory          |  2273|       50|svmRadial | 0.852| 0.086| 0.995|

but when I do scPredict(seurat_test, scpred), I get stuck at the following errors:

●  Matching reference with new dataset...
	 ─ 2000 features present in reference loadings
	 ─ 2000 features shared between reference and new dataset
	 ─ 100% of features in the reference are present in new dataset
●  Aligning new data to reference...
Harmony 1/20
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Harmony 2/20
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Harmony 3/20
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Harmony converged after 3 iterations
●  Classifying cells...
Error in `.rowNamesDF<-`(x, value = value) : invalid 'row.names' length
In addition: Warning message:
In method$prob(modelFit = modelFit, newdata = newdata, submodels = param) :
  kernlab class probability calculations failed; returning NAs

Any suggestion on how to proceed? I generated seurat_test and seurat_train splicing the original dataset using a 4-fold cross-validation method, scPred is a scpred object trained on seurat_train.

Cheers,
Carlo

sessionInfo()
R version 4.0.3 (2020-10-10)
Platform: x86_64-apple-darwin17.0 (64-bit)
Running under: macOS Big Sur 10.16

Matrix products: default
LAPACK: /Library/Frameworks/R.framework/Versions/4.0/Resources/lib/libRlapack.dylib

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] parallel  stats4    stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] pbapply_1.4-3               scPred_1.9.0                magrittr_2.0.1              doParallel_1.0.16           iterators_1.0.13           
 [6] foreach_1.5.1               SeuratObject_4.0.1          Seurat_4.0.1                SingleCellExperiment_1.12.0 SummarizedExperiment_1.20.0
[11] Biobase_2.50.0              GenomicRanges_1.42.0        GenomeInfoDb_1.26.7         IRanges_2.24.1              S4Vectors_0.28.1           
[16] BiocGenerics_0.36.1         MatrixGenerics_1.2.1        matrixStats_0.58.0          forcats_0.5.1               stringr_1.4.0              
[21] dplyr_1.0.6                 purrr_0.3.4                 readr_1.4.0                 tidyr_1.1.3                 tibble_3.1.2               
[26] ggplot2_3.3.3               tidyverse_1.3.1            

loaded via a namespace (and not attached):
  [1] readxl_1.3.1           backports_1.2.1        plyr_1.8.6             igraph_1.2.6           lazyeval_0.2.2         splines_4.0.3         
  [7] listenv_0.8.0          scattermore_0.7        digest_0.6.27          htmltools_0.5.1.1      fansi_0.4.2            tensor_1.5            
 [13] cluster_2.1.2          ROCR_1.0-11            recipes_0.1.16         globals_0.14.0         modelr_0.1.8           gower_0.2.2           
 [19] spatstat.sparse_2.0-0  colorspace_2.0-1       rvest_1.0.0            ggrepel_0.9.1          haven_2.4.1            xfun_0.23             
 [25] RCurl_1.98-1.3         crayon_1.4.1           jsonlite_1.7.2         spatstat.data_2.1-0    survival_3.2-11        zoo_1.8-9             
 [31] glue_1.4.2             polyclip_1.10-0        gtable_0.3.0           zlibbioc_1.36.0        XVector_0.30.0         ipred_0.9-11          
 [37] leiden_0.3.7           DelayedArray_0.16.3    kernlab_0.9-29         future.apply_1.7.0     abind_1.4-5            scales_1.1.1          
 [43] DBI_1.1.1              miniUI_0.1.1.1         Rcpp_1.0.6             viridisLite_0.4.0      xtable_1.8-4           reticulate_1.20       
 [49] spatstat.core_2.1-2    lava_1.6.9             prodlim_2019.11.13     htmlwidgets_1.5.3      httr_1.4.2             RColorBrewer_1.1-2    
 [55] ellipsis_0.3.2         ica_1.0-2              pkgconfig_2.0.3        nnet_7.3-16            uwot_0.1.10            dbplyr_2.1.1          
 [61] deldir_0.2-10          utf8_1.2.1             caret_6.0-88           tidyselect_1.1.1       rlang_0.4.11           reshape2_1.4.4        
 [67] later_1.2.0            munsell_0.5.0          cellranger_1.1.0       tools_4.0.3            cli_2.5.0              generics_0.1.0        
 [73] broom_0.7.6            ggridges_0.5.3         fastmap_1.1.0          goftest_1.2-2          ModelMetrics_1.2.2.2   knitr_1.33            
 [79] fs_1.5.0               fitdistrplus_1.1-3     RANN_2.6.1             future_1.21.0          nlme_3.1-152           mime_0.10             
 [85] xml2_1.3.2             rstudioapi_0.13        compiler_4.0.3         beeswarm_0.3.1         plotly_4.9.3           png_0.1-7             
 [91] spatstat.utils_2.1-0   reprex_2.0.0           stringi_1.6.2          highr_0.9              lattice_0.20-44        Matrix_1.3-3          
 [97] vctrs_0.3.8            pillar_1.6.1           lifecycle_1.0.0        spatstat.geom_2.1-0    lmtest_0.9-38          RcppAnnoy_0.0.18      
[103] bitops_1.0-7           data.table_1.14.0      cowplot_1.1.1          irlba_2.3.3            httpuv_1.6.1           patchwork_1.1.1       
[109] R6_2.5.0               promises_1.2.0.1       KernSmooth_2.23-20     gridExtra_2.3          vipor_0.4.5            parallelly_1.25.0     
[115] codetools_0.2-18       MASS_7.3-54            assertthat_0.2.1       withr_2.4.2            sctransform_0.3.2      GenomeInfoDbData_1.2.4
[121] harmony_1.0            mgcv_1.8-35            hms_1.0.0              grid_4.0.3             rpart_4.1-15           timeDate_3043.102     
[127] class_7.3-19           Rtsne_0.15             pROC_1.17.0.1          shiny_1.6.0            lubridate_1.7.10       tinytex_0.31          
[133] ggbeeswarm_0.6.0   

[nkarmi1]

Dear reader,

I ran into a similar problem.

Seurat object:

Trained scPred object:

After running SeuratObject <- scPredict(SeuratObject, scPredObject)
I get the following error:

Could you please provide a suggestion on how to continue the scPred script?
Thank you in advance.

Kind regards,
Naomi Karmi

[wenjie2wang]

Encountered the same issue. It seems that a simple solution is to (fork this repository and) comment out those rownames() commands in scPred::scPredict(). Alternatively, it is possible to modify the scPred::scPredict() directly as follows:

scPredict2 <- scPred::scPredict
b <- lapply(body(scPredict2), function(a) {
    gsub("(rownames\\(.*\\) <- .*)(;?\\n?)",
         "message('Try to set the row names.'); try(\\1)\\2", deparse(a))
})
body(scPredict2) <- parse(text = c(do.call(c, b), "}"))
environment(scPredict2) <- environment(scPred::scPredict)
assignInNamespace("scPredict", scPredict2, ns = asNamespace("scPred"))

## example
library(magrittr)
library(Seurat)
#> Attaching SeuratObject
#> Attaching sp
library(scPred)

reference <- scPred::pbmc_1
query <- scPred::pbmc_2
reference <- reference %>%
  NormalizeData() %>%
  FindVariableFeatures() %>%
  ScaleData() %>%
  RunPCA() %>%
  RunUMAP(dims = 1:30)
#> Centering and scaling data matrix
#> PC_ 1 
#> Positive:  MALAT1, RPS12, RPL13, RPS18, RPS4X, LTB, IL32, IL7R, EEF1A1, RPL10 
#>     NOSIP, RPS2, CD2, CD7, CCR7, CD27, CD69, MAL, TSHZ2, ITM2A 
#>     MYC, TRAT1, CD247, FHIT, GZMM, SH3YL1, TXK, AQP3, TESPA1, PIM1 
#> Negative:  CST3, FCN1, LYZ, LST1, CSTA, S100A9, SERPINA1, CD68, AIF1, SPI1 
#>     CFD, MNDA, S100A8, VCAN, TYROBP, KLF4, CLEC7A, CTSS, PSAP, S100A12 
#>     FCER1G, MS4A6A, TYMP, FTL, NCF2, FGL2, CYBB, IFI30, IGSF6, CD14 
#> PC_ 2 
#> Positive:  NKG7, CST7, GNLY, GZMA, GZMB, FGFBP2, PRF1, CCL5, B2M, CTSW 
#>     KLRD1, GZMH, KLRF1, CCL4, HOPX, SPON2, CLIC3, CMC1, KLRB1, GZMM 
#>     FCGR3A, IFITM2, TMSB4X, EFHD2, MATK, ITGB2, MYO1F, PFN1, SRGN, CD247 
#> Negative:  RPL13, RPS12, RPS18, LTB, RPS2, EEF1A1, RPL10, RPLP1, RPS4X, TPT1 
#>     MS4A1, CD79A, HLA-DRA, BANK1, MEF2C, LINC00926, HLA-DQB1, HLA-DQA1, VPREB3, CD79B 
#>     NCF1, IL7R, CCR7, BIRC3, TNFRSF13C, MARCH1, RALGPS2, TCL1A, HLA-DMB, CD22 
#> PC_ 3 
#> Positive:  CD79A, MS4A1, CD79B, HLA-DQA1, CD74, BANK1, HLA-DRA, LINC00926, HLA-DPB1, HLA-DPA1 
#>     HLA-DQB1, HLA-DRB1, MEF2C, VPREB3, TCL1A, RALGPS2, HLA-DMB, JCHAIN, BCL11A, HLA-DMA 
#>     BLK, TCF4, CD22, ADAM28, SPIB, TNFRSF13C, MARCH1, IGLL5, LINC01781, IRF8 
#> Negative:  IL7R, TPT1, NOSIP, RPS12, VIM, IL32, MAL, S100A12, S100A9, S100A8 
#>     VCAN, AIF1, ANXA1, TRAT1, RPL13, S100A10, FCN1, CD14, CCR7, CSTA 
#>     TSHZ2, LYZ, S100A6, RGCC, RPS4X, RPL10, TNFAIP3, CD2, RBP7, RP11-1143G9.4 
#> PC_ 4 
#> Positive:  S100A12, VCAN, MS4A6A, S100A8, CD14, MGST1, CD36, RNASE6, RP11-1143G9.4, PLBD1 
#>     CXCL8, S100A9, CSF3R, RBP7, GPX1, MNDA, QPCT, LYZ, CLEC4E, ALDH2 
#>     LGALS2, CYP1B1, FOLR3, KCTD12, GRN, FPR1, PTPRE, RETN, SERPINF1, RP11-160E2.6 
#> Negative:  CDKN1C, HES4, TCF7L2, MS4A7, CKB, SMIM25, VMO1, HMOX1, LYPD2, FCGR3A 
#>     CXCL16, ICAM4, SIGLEC10, CSF1R, IFITM3, PILRA, MEG3, LILRA5, LRRC25, ADGRE1 
#>     MTSS1, LILRB2, ADGRE2, PECAM1, C1QA, ZNF703, TNFRSF8, IFI30, WARS, SPI1 
#> PC_ 5 
#> Positive:  SERPINF1, LILRA4, LRRC26, PLD4, PPP1R14B, TPM2, ITM2C, SCN9A, PTCRA, MZB1 
#>     SCT, LINC00996, DNASE1L3, CLEC4C, IRF7, UGCG, C1orf186, GAS6, TCF4, LILRB4 
#>     RP11-117D22.2, CCDC50, DERL3, MYBL2, JCHAIN, APP, IL3RA, ZFAT, SPIB, IRF8 
#> Negative:  MS4A1, CD79A, LINC00926, CD79B, VCAN, S100A12, VPREB3, BANK1, S100A8, FCER2 
#>     CD22, ARHGAP24, TNFRSF13C, RALGPS2, LY86, CD14, MNDA, FCRL1, LINC01781, HVCN1 
#>     S100A9, BLK, ADAM28, PLBD1, CSF3R, RP11-1143G9.4, CD24, CLEC4E, RBP7, CYP1B1
#> Warning: The default method for RunUMAP has changed from calling Python UMAP via reticulate to the R-native UWOT using the cosine metric
#> To use Python UMAP via reticulate, set umap.method to 'umap-learn' and metric to 'correlation'
#> This message will be shown once per session
#> 22:58:36 UMAP embedding parameters a = 0.9922 b = 1.112
#> 22:58:36 Read 3500 rows and found 30 numeric columns
#> 22:58:36 Using Annoy for neighbor search, n_neighbors = 30
#> 22:58:36 Building Annoy index with metric = cosine, n_trees = 50
#> 0%   10   20   30   40   50   60   70   80   90   100%
#> [----|----|----|----|----|----|----|----|----|----|
#> **************************************************|
#> 22:58:36 Writing NN index file to temp file /var/folders/qk/8d34qqss0538pp8mm6y8g0840000gn/T//RtmpC4AaeV/file37c178ef3c10
#> 22:58:36 Searching Annoy index using 1 thread, search_k = 3000
#> 22:58:37 Annoy recall = 100%
#> 22:58:37 Commencing smooth kNN distance calibration using 1 thread
#> 22:58:37 Initializing from normalized Laplacian + noise
#> 22:58:38 Commencing optimization for 500 epochs, with 158046 positive edges
#> 22:58:43 Optimization finished
reference <- getFeatureSpace(reference, "cell_type")
#> ●  Extracting feature space for each cell type...
#> DONE!
reference <- trainModel(reference)
#> ●  Training models for each cell type...
#> Loading required package: ggplot2
#> Loading required package: lattice
#> maximum number of iterations reached 0.0001163544 -0.0001154305DONE!
query <- NormalizeData(query)
query <- scPredict(query, reference)
#> ●  Matching reference with new dataset...
#>   ─ 2000 features present in reference loadings
#>   ─ 2000 features shared between reference and new dataset
#>   ─ 100% of features in the reference are present in new dataset
#> ●  Aligning new data to reference...
#> Harmony 1/20
#> Harmony 2/20
#> Harmony 3/20
#> Harmony 4/20
#> Harmony 5/20
#> Harmony 6/20
#> Harmony 7/20
#> Harmony 8/20
#> Harmony 9/20
#> Harmony 10/20
#> Harmony converged after 10 iterations
#> ●  Classifying cells...
#> Try to set the row names.
#> Try to set the row names.
#> Try to set the row names.
#> Try to set the row names.
#> Try to set the row names.
#> Try to set the row names.
#> Try to set the row names.
#> Try to set the row names.
#> Try to set the row names.
#> Try to set the row names.
#> DONE!
crossTab(query, "cell_type", "scpred_prediction")
#>             B cell CD4 T cell CD8 T cell cDC cMono ncMono NK cell pDC
#> B cell         250          0          0   0     3      0       0   0
#> CD4 T cell       0       1178         25   0     3      0       0   0
#> CD8 T cell       0         47        602   0     1      0      39   0
#> cDC              0          0          0  20    14      0       0   1
#> cMono            0          0          0   0   281      0       0   0
#> ncMono           0          1          0   0     1    119       0   0
#> NK cell          0          0         11   0     0      0     230   0
#> pDC              0          0          0   0     0      0       0  11
#> Plasma cell      0          0          0   0     0      0       0   0
#> unassigned       0         12         38   0   106      0       1   0
#>             Plasma cell
#> B cell                0
#> CD4 T cell            0
#> CD8 T cell            0
#> cDC                   0
#> cMono                 0
#> ncMono                0
#> NK cell               0
#> pDC                   0
#> Plasma cell           6
#> unassigned            0

^{Created on 2022-06-05 by the reprex package (v2.0.1)}

Notice that I added the message "Try to set the row names." on purpose to confirm that the function scPred::predict() had been modified to set the row names optionally.

[gijreyneveld]

Hi Naomi,

Looking at the Celltypes, I think I am using the same dataset as a reference. Were you able to solve this issue yet, as the fork suggestion is not working for me.

Best,

IJsbrand