Different p_val_adj manually and Seurat #10063
-
|
Hi, I’m experiencing an issue with the p_val_adj. I noticed that the adjusted p-values returned by FindMarkers() are not the same as those I obtain when applying the correction manually using p.adjust(p_raw, method = "fdr"). I ran the analysis using all genes (min.pct = 0), but I still get different adjusted p-values (p_val_adj) between the two approaches. In fact, I obtained more significant genes and DARs when doing the correction manually than when using FindMarkers(). How is this possible? According to the FindMarkers() manual, FDR-based correction methods are not recommended because Seurat pre-filters genes using parameters such as logfc.threshold = 0.25 or min.pct = 0.1, which reduces the number of tests performed. This reduced test set should still be accounted for, but FDR-based methods cannot properly handle this. My question is: does FindMarkers() filter the genes first (by min.pct = 0.1 and logfc.threshold = 0.25) before calculating the adjusted p-values (p_val_adj)? Thanks! |
Beta Was this translation helpful? Give feedback.
Answered by
samuel-marsh
Aug 29, 2025
Replies: 2 comments 3 replies
-
|
Hi, So first just as note the results would not match when using code you provided because Seurat doesn't use FDR correction but bonferroni. Second, as has been discussed on GitHub a bunch, Seurat corrects vs. all genes present in object which you can see in raw code here: seurat/R/differential_expression.R Lines 627 to 631 in e44cb2c Best, |
Beta Was this translation helpful? Give feedback.
2 replies
-
|
Thank you so much for your answers! However, when I change the method in Seurat to "BH", the results are the same as those I get using Bonferroni, and they are different from what I get when I do it manually. What could be the reason for this? |
Beta Was this translation helpful? Give feedback.
-
|
Can you please provide full code that you are running? |
Beta Was this translation helpful? Give feedback.
Answer selected by
mluciarr
-
|
Here you go I did the analysis using the default filtering and then removing the filtering and applying the filtering afterwards. |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
Hi, So again I think issue is again that you are using FDR adjustment and not providing all comparisons when using Here is reprex showing identical results manually vs Seurat using pbmc3k object. Best, library(tidyverse)
library(Seurat)
#> Loading required package: SeuratObject
#> Loading required package: sp
#>
#> Attaching package: 'SeuratObject'
#> The following objects are masked from 'package:base':
#>
#> intersect, t
library(scCustomize)
#> scCustomize v3.1.0
#> If you find the scCustomize useful please cite.
#> See 'samuel-marsh.github.io/scCustomize/articles/FAQ.html' for citation info.
pbmc <- pbmc3k.SeuratData::pbmc3k.final
pbmc <- UpdateSeuratObject(pbmc)
#> Validating object structure
#> Updating object slots
#> Ensuring keys are in the proper structure
#> Updating matrix keys for DimReduc 'pca'
#> Updating matrix keys for DimReduc 'umap'
#> Warning: Assay RNA changing from Assay to Assay
#> Warning: Graph RNA_nn changing from Graph to Graph
#> Warning: Graph RNA_snn changing from Graph to Graph
#> Warning: DimReduc pca changing from DimReduc to DimReduc
#> Warning: DimReduc umap changing from DimReduc to DimReduc
#> Ensuring keys are in the proper structure
#> Ensuring feature names don't have underscores or pipes
#> Updating slots in RNA
#> Updating slots in RNA_nn
#> Setting default assay of RNA_nn to RNA
#> Updating slots in RNA_snn
#> Setting default assay of RNA_snn to RNA
#> Updating slots in pca
#> Updating slots in umap
#> Setting umap DimReduc to global
#> Setting assay used for NormalizeData.RNA to RNA
#> Setting assay used for FindVariableFeatures.RNA to RNA
#> Setting assay used for ScaleData.RNA to RNA
#> Setting assay used for RunPCA.RNA to RNA
#> Setting assay used for JackStraw.RNA.pca to RNA
#> No assay information could be found for ScoreJackStraw
#> Warning: Adding a command log without an assay associated with it
#> Setting assay used for FindNeighbors.RNA.pca to RNA
#> No assay information could be found for FindClusters
#> Warning: Adding a command log without an assay associated with it
#> Setting assay used for RunUMAP.RNA.pca to RNA
#> Validating object structure for Assay 'RNA'
#> Validating object structure for Graph 'RNA_nn'
#> Validating object structure for Graph 'RNA_snn'
#> Validating object structure for DimReduc 'pca'
#> Validating object structure for DimReduc 'umap'
#> Object representation is consistent with the most current Seurat version
seurat_markers <- FindMarkers(object = pbmc, ident.1 = "B", ident.2 = "NK")
seurat_padj <- seurat_markers$p_val_adj
manual_padj <- p.adjust(p = seurat_markers$p_val, method = "bonferroni", n = nrow(pbmc))
identical(manual_padj, seurat_padj)
#> [1] TRUECreated on 2025-09-04 with reprex v2.1.1
|
Beta Was this translation helpful? Give feedback.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hi,
So first just as note the results would not match when using code you provided because Seurat doesn't use FDR correction but bonferroni. Second, as has been discussed on GitHub a bunch, Seurat corrects vs. all genes present in object which you can see in raw code here:
seurat/R/differential_expression.R
Lines 627 to 631 in e44cb2c
Best,
Sam