small sample sizes failing at batch correction #194
Description
CCA and RPCA k.weight default is 100. If a sample has < 100 cells then integration by these methods fails.
Perhaps need way to override the k.weight default when there are samples with very few QC'd cells?
Example of error below where 3 samples in the data set had 56 -79 cells:
ERROR ~ Error executing process > 'GEX_EXQC:BATCH_CORRECT_CCA (PBMC-Blood-Cytology-CTC-Biopsy-Autopsy-AutopsyNormal)'
Caused by:
Process `GEX_EXQC:BATCH_CORRECT_CCA (PBMC-Blood-Cytology-CTC-Biopsy-Autopsy-AutopsyNormal)` terminated with an error exit status (1)
Command executed:
Rscript -e 'rmarkdown::render("batch_correction_cca.Rmd",
params=list(gid="PBMC-Blood-Cytology-CTC-Biopsy-Autopsy-AutopsyNormal",
mergedObj="PBMC-Blood-Cytology-CTC-Biopsy-Autopsy-AutopsyNormal_seurat_merged.rds",
species="hg38",
npcs="50",
vars_to_regress="NULL",
resolution_list="0.1,0.2,0.3,0.5,0.6,0.8,1",
Rlib_dir="/data/CCBR_Pipeliner/db/PipeDB/Rlibrary_4.3_scRNA_RHEL8",
Rpkg_config="Rpack.config",
scRNA_functions="scRNA_functions.R",
testing="N"),
output_file = "PBMC-Blood-Cytology-CTC-Biopsy-Autopsy-AutopsyNormal_batch_correction_cca.html")'
Command exit status:
1
Command output:
1/8
2/8 [prep_args]
3/8
4/8 [handle_pkg]
5/8
6/8 [processing]
Command error:
|
|=================================== | 50%
|
|==================================================== | 75%
|
|======================================================================| 100%
|
| | 0%
|
|================== | 25%
|
|=================================== | 50%
|
|==================================================== | 75%
|
|======================================================================| 100%
|
| | 0%
|
|================== | 25%
|
|=================================== | 50%
|
|==================================================== | 75%
|
|======================================================================| 100%
|
| | 0%
|
|================== | 25%
|
|=================================== | 50%
|
|==================================================== | 75%
|
|======================================================================| 100%
Quitting from lines at lines 66-79 [processing] (batch_correction_cca.Rmd)
Error:
! k.weight (100) is set larger than the number of cells in the smallest object (56). Please choose a smaller k.weight.
Backtrace:
1. global MAIN_BATCH_CORRECTION(...)
2. Seurat::IntegrateLayers(...)
3. Seurat (local) method(...)
5. Seurat:::IntegrateEmbeddings.IntegrationAnchorSet(...)
6. Seurat:::ValidateParams_IntegrateEmbeddings_IntegrationAnchors(...)
Execution halted