Error in design

[desmodus1984]

Hi,
I am trying to decide whether it is worth to use lfcShrink and I am trying to use mirrorCheck.
I created a DESeq object with a metadata and a count matrix.

miRNA.Dec22.cetaceans <- DESeqDataSetFromMatrix(countData = subset.miRNA.Dec.22.cetaceans,
                                             colData = subset.metadata.Dec.22,
                                             design = ~ sp)

Then, I used sva to find surrogate variables and I added them to the design.

mod <- model.matrix(~ sp, colData(miRNA.Dec22.cetaceans))
mod0 <- model.matrix(~ 1, colData(miRNA.Dec22.cetaceans))

svseq <- svaseq(dat, mod, mod0,B=10)
miRNA.Dec22.cetaceans_sva <- miRNA.Dec22.cetaceans
miRNA.Dec22.cetaceans_sva$SV1 <- svseq$sv[,1]
miRNA.Dec22.cetaceans_sva$SV2 <- svseq$sv[,2]
miRNA.Dec22.cetaceans_sva$SV3 <- svseq$sv[,3]
design(miRNA.Dec22.cetaceans_sva) <- ~ SV1 + SV2 + SV3 + sp

When I did lfcshrink, the LFC values shrunk and the miRNAs that were previously meeting enough LFC (abs()>=0.5), didn't any more.
Before
baseMean log2FoldChange lfcSE stat pvalue padj

hsa-miR-122-5p 217.373 -3.71737 1.83422 29.9837 3.95900e-05 0.01621210
hsa-miR-1-3p 363.626 -3.16970 2.17947 33.0295 1.03496e-05 0.00847631

After
baseMean log2FoldChange lfcSE pvalue padj

hsa-miR-122-5p 217.373 1.10653e-07 0.00144269 3.95900e-05 0.01621210
hsa-miR-1-3p 363.626 -2.85928e-07 0.00144270 1.03496e-05 0.00847631

I tried running mirrorCheck and it failed:

run_DESeq_all_contrasts(  miRNA.Dec22.cetaceans_sva,
+                           folder = results,
+                            mode = "apeglm",
+                            condition="sp",
+                            rowname2symbol = NULL,
+                            heatmap.annotation.col = NULL,
+                            p.cutoff = 0.1,
+                            fc.cutoff = 0.5,
+                            top.n = 50,
+                            useDingbats = F,
+                            print.all = F
+ )

Error in file.path(folder) :
cannot coerce type 'closure' to vector of type 'character'

Any reason, why this happened?
sp, isn't recognized.

run_DESeq_all_contrasts(  miRNA.Dec22.cetaceans_sva,
+                           folder = results,
+                            mode = "apeglm",
+                            condition=sp,
+                            rowname2symbol = NULL,
+                            heatmap.annotation.col = NULL,
+                            p.cutoff = 0.1,
+                            fc.cutoff = 0.5,
+                            top.n = 50,
+                            useDingbats = F,
+                            print.all = F
+ )

Error: object 'sp' not found

[kescull]

Hi, sorry for the delay in replying. Happy holidays and happy new year!
Both the folder and condition parameters need to be character strings, i.e. probably inside quote marks. In the case of your folder parameter, you have probably already assigned a DESeqResults object to this variable, since you say you have tried DESeq2 without lfcShrink. (By the way, you can also do this using mirrorCheck, by setting mode = "none".) So your code would pass your previous results to run_DESeq_all_contrasts, rather than the character string it is looking for. Thanks for reporting the error; I tried to write it so that it will exist gracefully if a folder of the given name doesn't exist, but I didn't realise it wouldn't fail appropriately in a case like yours. Long story short, you probably want folder="results". Also, don't forget to make a folder called results in your working directory before running the function.
Secondly, you probably just need to put sp in quote marks. condition="sp". Again, currently your code will look for a variable called sp, rather than passing the character string "sp". So unless for some reason you had already run sp="sp", your code won't do what you want.
Hope that's a simple fix for you!