edist calculation difference in edist and etest

[rootze]

Hello Developer,

First, thank you for providing this great resource. I greatly appreciate the effort that went into developing and maintaining this tool.

I ran into an issue that has me quite confused. Specifically, I ran the code available in testthat

Here is the part of the code I'm referring to:

expression_matrix <- Seurat::ReadMtx(
  paste0(path, "/matrix.mtx"),
  paste0(path, "/barcodes.csv"),
  paste0(path, "/genes.csv"),
  cell.column = 1,
  feature.column = 2,
  cell.sep = ",",
  feature.sep = ",",
  skip.cell = 1,
  skip.feature = 1,
latent <- read.csv(paste0(path, "seurat_pca.csv"), row.names = 1)
metadata <- read.csv(paste0(path, "seurat_metadata.csv"),
                     row.names = 1)
colnames(expression_matrix) <- rownames(metadata)
seurat_object <- Seurat::CreateSeuratObject(counts = expression_matrix,
                                             meta.data = metadata)
seurat_object[["pca"]] <- Seurat::CreateDimReducObject(embeddings = as.matrix(latent),
                                               key = "pca_",
                                               assay = Seurat::DefaultAssay(seurat_object))

df <- edist(seurat_object, groupby = "perturbation", reduction = "pca",
            sample_correction = FALSE, verbose = TRUE)
#
df
# > df
#            RUNX2_2   control
# ZAP70_1  0.1849568 0.2478410
# ZAP70_2  0.1320942 0.3173424
# LCK_1    0.1971550 0.2685310

#
et <- etest(seurat_object = seurat_object, groupby = "perturbation",
            control = "control", reduction = "pca")
#
et
# > et
#                 pval        edist
# ZAP70_1  0.000999001 2.478410e+01
# ZAP70_2  0.000999001 3.173424e+01

I noticed that the edist value calculated using the edist function differs significantly from the value obtained using the etest function for the comparison between ZAP70_1 and control.

For example:

In the edist dataframe, the value is 0.2478410.
In the output from etest, the corresponding value is 2.478410e+01.

I'm having trouble understanding why there is such a difference between these two calculations, and what might be causing this discrepancy. Could you provide some clarification or point me in the right direction?
Thank you!

[tessadgreen]

Apologies that we haven't had the chance to investigate this—at the moment, the re-implementation at pertpy is being more actively maintained, so I'd recommend trying that instead.