Dear community,

i have an issue with the write.fit function which I use to create a .txt of a limma object because the summary(results) differs from the values in the .txt independent of the 'adjust.method' or the 'method' I choose. (tried different options in decideTests as well as write.fit). write.fit doesn't respond to any change. 

Compared to summary(result) the number of significant genes in contrasts can be reduced by a hundred genes when a file is created with write.fit.  

.....

fit2 <- eBayes(fit2)

results <- decideTests(fit2, method = "global", adjust.method = "BH")

write.fit(fit2, results, "limma_results2.txt", adjust="none", method ="separate")

I'm looking forward to your comments!

Tillmann

R version 3.5.0 (2018-04-23)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS Sierra 10.12.6

Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.5/Resources/lib/libRlapack.dylib

locale:
[1] de_DE.UTF-8/de_DE.UTF-8/de_DE.UTF-8/C/de_DE.UTF-8/de_DE.UTF-8

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] data.table_1.11.4    limma_3.36.1         qvalue_2.12.0        BiocInstaller_1.30.0

loaded via a namespace (and not attached):
 [1] Rcpp_0.12.17     grid_3.5.0       plyr_1.8.4       gtable_0.2.0     magrittr_1.5    
 [6] scales_0.5.0     ggplot2_2.2.1    pillar_1.2.3     stringi_1.2.3    rlang_0.2.1     
[11] reshape2_1.4.3   lazyeval_0.2.1   splines_3.5.0    tools_3.5.0      stringr_1.3.1   
[16] munsell_0.5.0    compiler_3.5.0   colorspace_1.3-2 tibble_1.4.2  

Works fine for me.

library(limma)
g <- gl(3, 2)
y <- matrix(rnorm(length(g) * 10000), ncol=6)

design <- model.matrix(~g)
fit <- lmFit(y, design)
fit <- eBayes(fit)

results <- decideTests(fit, method = "global", adjust.method = "BH")
write.fit(fit, results, "limma_results1.txt", adjust="none", method = "separate")
write.fit(fit, results, "limma_results2.txt", adjust="BH", method ="separate")

first <- read.delim("limma_results1.txt", header=TRUE, check.names=FALSE)
colnames(first)
##  [1] "A"                   "Coef.(Intercept)"    "Coef.g2"            
##  [4] "Coef.g3"             "t.(Intercept)"       "t.g2"               
##  [7] "t.g3"                "p.value.(Intercept)" "p.value.g2"         
## [10] "p.value.g3"          "F"                   "F.p.value"          
## [13] "Res.(Intercept)"     "Res.g2"              "Res.g3"             
second <- read.delim("limma_results2.txt", header=TRUE, check.names=FALSE)
##  [1] "A"                       "Coef.(Intercept)"       
##  [3] "Coef.g2"                 "Coef.g3"                
##  [5] "t.(Intercept)"           "t.g2"                   
##  [7] "t.g3"                    "p.value.(Intercept)"    
##  [9] "p.value.g2"              "p.value.g3"             
## [11] "p.value.adj.(Intercept)" "p.value.adj.g2"         
## [13] "p.value.adj.g3"          "F"                      
## [15] "F.p.value"               "Res.(Intercept)"        
## [17] "Res.g2"                  "Res.g3"

You can see that if you turn adjust off in write.fit, you don't even get the adjusted p-value columns in first. And obviously, you will get different adjustments between write.fit and decideTests if you change method=.