Question: More significant genes with alpha 0.05 than 0.1 DESeq2
gravatar for frene
6 months ago by
frene0 wrote:

Hello everybody,

As I can read on the DESEQ2 manual, if we want an other p-adj diferent than 0.1 we have to select it by changing the alpha value. When I change it in the result formula:

alpha <- 0.05
res_0.05_filt <- results(dds1,contrast=c("condition","WT", "OE"),alpha=alpha)

Why I obtain more significant genes with p-adj 0.05 than 0.1? Why the mean count change in both cases? It is correct to do the results with 0.1 and them filter by 0.05?

here is the DESeq2 exit:

**> summary(res1)

out of 24732 with nonzero total read count
adjusted p-value < 0.1
LFC > 0 (up)       : 17, 0.069%
LFC < 0 (down)     : 24, 0.097%
outliers [1]       : 0, 0%
low counts [2]     : 6672, 27%
(mean count < 17)
[1] see 'cooksCutoff' argument of ?results
[2] see 'independentFiltering' argument of ?results

> table(res1$padj<=0.05)

18050    10 
> summary(res_0.05_filt)

out of 24732 with nonzero total read count
adjusted p-value < 0.05
LFC > 0 (up)       : 8, 0.032%
LFC < 0 (down)     : 13, 0.053%
outliers [1]       : 0, 0%
low counts [2]     : 16679, 67%
(mean count < 239)
[1] see 'cooksCutoff' argument of ?results
[2] see 'independentFiltering' argument of ?results

> table(res_0.05_filt$padj<=0.05)

 8032    21**

Here is the sessioninfo:

> sessionInfo()
R version 3.5.1 (2018-07-02)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows >= 8 x64 (build 9200)

Matrix products: default

[1] LC_COLLATE=Spanish_Spain.1252  LC_CTYPE=Spanish_Spain.1252    LC_MONETARY=Spanish_Spain.1252 LC_NUMERIC=C                  
[5] LC_TIME=Spanish_Spain.1252    

attached base packages:
 [1] grid      parallel  stats4    stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] ggbiplot_0.55               scales_1.0.0                plyr_1.8.4                  corrplot_0.84               RColorBrewer_1.1-2         
 [6] pheatmap_1.0.12             sva_3.30.1                  genefilter_1.64.0           mgcv_1.8-28                 nlme_3.1-137               
[11] gplots_3.0.1.1              ggplot2_3.1.0               limma_3.38.3                DESeq2_1.22.2               SummarizedExperiment_1.12.0
[16] DelayedArray_0.8.0          BiocParallel_1.16.6         matrixStats_0.54.0          Biobase_2.42.0              GenomicRanges_1.34.0       
[21] GenomeInfoDb_1.18.2         IRanges_2.16.0              S4Vectors_0.20.1            BiocGenerics_0.28.0        

loaded via a namespace (and not attached):
 [1] bit64_0.9-7            splines_3.5.1          gtools_3.8.1           Formula_1.2-3          latticeExtra_0.6-28    blob_1.1.1            
 [7] GenomeInfoDbData_1.2.0 pillar_1.3.1           RSQLite_2.1.1          backports_1.1.3        lattice_0.20-38        digest_0.6.18         
[13] XVector_0.22.0         checkmate_1.9.1        colorspace_1.4-1       htmltools_0.3.6        Matrix_1.2-17          XML_3.98-1.19         
[19] pkgconfig_2.0.2        zlibbioc_1.28.0        xtable_1.8-3           gdata_2.18.0           htmlTable_1.13.1       tibble_2.1.1          
[25] annotate_1.60.1        withr_2.1.2            nnet_7.3-12            lazyeval_0.2.2         survival_2.43-3        magrittr_1.5          
[31] crayon_1.3.4           memoise_1.1.0          MASS_7.3-51.1          foreign_0.8-71         tools_3.5.1            data.table_1.12.0     
[37] stringr_1.4.0          munsell_0.5.0          locfit_1.5-9.1         cluster_2.0.7-1        AnnotationDbi_1.44.0   compiler_3.5.1        
[43] caTools_1.17.1.2       rlang_0.3.2            RCurl_1.95-4.12        rstudioapi_0.10        htmlwidgets_1.3        labeling_0.3          
[49] bitops_1.0-6           base64enc_0.1-3        gtable_0.3.0           DBI_1.0.0              gridExtra_2.3          knitr_1.22            
[55] bit_1.1-14             Hmisc_4.2-0            KernSmooth_2.23-15     stringi_1.4.3          Rcpp_1.0.1             geneplotter_1.60.0    
[61] rpart_4.1-13           acepack_1.4.1          xfun_0.5              
deseq2 • 107 views
ADD COMMENTlink modified 6 months ago by Michael Love25k • written 6 months ago by frene0
Answer: More significant genes with alpha 0.05 than 0.1 DESeq2
gravatar for Michael Love
6 months ago by
Michael Love25k
United States
Michael Love25k wrote:

This is just what you would expect from the documentation. The alpha you provide to results() informs the independent filtering, and giving it the alpha you intend to use will give you the most rejections.

ADD COMMENTlink written 6 months ago by Michael Love25k

Thanks Michel,

I saw a problem because with alpha 0.1 I obtained 10 genes and with 0.05, 21 genes, but it's true that the low counts rejection number of genes is greater with 0.05.

Another related question to de independent filtering. This is probably because of my limited knowledge , but If you turn it off with independentFiltering=FALSE, you get the p-adj values for all the genes but the values are higher than if you do with the independentFiltering=TRUE (default).
is there any way to make the independent Filtering with a basemean value defined by the user?

thanks a lot.

ADD REPLYlink written 6 months ago by frene0

No it’s an automatic procedure (see paper or vignettte).

You can turn it off and filter the results table and then recompute adjusted pvalues with p.adjust().

ADD REPLYlink written 6 months ago by Michael Love25k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 317 users visited in the last hour