search

wgmao / PLIER

Pathway-Level Information Extractor (PLIER): a generative model for gene expression data
Other
52 stars 9 forks source link

Error in PLIER while performing pseudobulk analysis with single-cell data (Error in svd(Y) : infinite or missing values in 'x') #18

Open recervictory opened 1 day ago

recervictory commented 1 day ago

Hello,

I am attempting to use the PLIER package for pseudobulk analysis of single-cell RNA sequencing data. I have generated pseudobulk data using the PseudobulkExpression function from Seurat and passed it to the PLIER function. However, I encounter an error during the SVD computation:

` pseudobulk_data <- PseudobulkExpression( object = seuratObject, group.by = c("sample", "cell_type"), # Grouping by sample and cell type assays = "RNA", # Assay to use features = NULL, # Use all features method = "average", # Aggregation method normalization.method = "LogNormalize",# Normalization method scale.factor = 10000, # Scale factor for normalization verbose = TRUE, # Print progress layer = "data" )

plierResult <- PLIER(pseudobulk_data, allPaths)

ERROR Selecting common genes: 6078 Removing 12 pathways with too few genes Computing SVD Error in svd(Y) : infinite or missing values in 'x'

`

wgmao commented 1 day ago

Could you check if there are any NA or infinite values in pseudobulk_data? If the data looks good, the issues might be due to some genes having constant values (e.g., all zeros) across all cells. By default, PLIER applies the z-score transformation on genes, which can result in constant gens being converted to NA values.