6 groups to compare DESeq2 (Wald vs LRT)
0
0
Entering edit mode
Rocio • 0
@2d75b62a
Last seen 1 day ago
United Kingdom

I have a comparison of 6 groups: healthy, disease, longer_disease, treatment1, treatment2, treatment3. I have previously used LRT. I can consider longer_disease as the group to compare to, and I am only interested in comparing the rest of groups to that one, not in every individual comparison.

I have used LRT before in other dataset and I did not have this observation. On re-running (and re-reading) about it, I get the same number of DEGs in each individual comparison (e.g. treatment1_vs_longer_disease or treatment2_vs_longer_disease).

Should I use Wald... how?

Thank you! Help!

Code should be placed in three backticks as shown below


# include your problematic code here with any corresponding output 
dds <- DESeqDataSetFromMatrix(countData = cts,
                              colData = coldata,
                              design = ~ group)

#where group is the factor of 5 variables 
dds$group <- relevel(dds$group, ref = "longer_disease") # to set it as my baseline (this makes sense biololgically and for our question)

keep <- rowSums(counts(dds) >= 10) >= 5 # since I have a minimum of 5 biological replicates per group
dds <- dds[keep,]

dds <- DESeq(dds, test="LRT", reduced=~1) # considering our design is comparing between groups 
res <- results(dds)

resultsNames(dds)
[1] "Intercept"             "group_treatment1_vs_longer_disease"    
[3] "group_healthy_vs_longer_disease" "group_treatment2_vs_longer_disease"    
[5] "group_treatment2_vs_longer_disease"  "group_disease_vs_longer_disease"

# please also include the results of running the following in an R session 

sessionInfo( )

R version 4.4.1 (2024-06-14)
Platform: x86_64-apple-darwin20
Running under: macOS Monterey 12.6.2

Matrix products: default
BLAS:   /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib 
LAPACK: /Library/Frameworks/R.framework/Versions/4.4-x86_64/Resources/lib/libRlapack.dylib;  LAPACK version 3.12.0

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

time zone: Europe/London
tzcode source: internal

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

other attached packages:
 [1] lubridate_1.9.4            
 [2] forcats_1.0.0              
 [3] stringr_1.5.1              
 [4] dplyr_1.1.4                
 [5] purrr_1.0.4                
 [6] readr_2.1.5                
 [7] tidyr_1.3.1                
 [8] tibble_3.2.1               
 [9] ggplot2_3.5.2              
[10] tidyverse_2.0.0            
[11] VennDiagram_1.7.3          
[12] futile.logger_1.4.3        
[13] pheatmap_1.0.12            
[14] DESeq2_1.44.0              
[15] SummarizedExperiment_1.34.0
[16] Biobase_2.64.0             
[17] MatrixGenerics_1.16.0      
[18] matrixStats_1.5.0          
[19] GenomicRanges_1.56.2       
[20] GenomeInfoDb_1.40.1        
[21] IRanges_2.38.1             
[22] S4Vectors_0.42.1           
[23] BiocGenerics_0.50.0        

loaded via a namespace (and not attached):
 [1] gtable_0.3.6            lattice_0.22-7         
 [3] tzdb_0.5.0              vctrs_0.6.5            
 [5] tools_4.4.1             generics_0.1.4         
 [7] parallel_4.4.1          pkgconfig_2.0.3        
 [9] Matrix_1.7-3            RColorBrewer_1.1-3     
[11] lifecycle_1.0.4         GenomeInfoDbData_1.2.12
[13] compiler_4.4.1          farver_2.1.2           
[15] codetools_0.2-20        pillar_1.10.2          
[17] crayon_1.5.3            BiocParallel_1.38.0    
[19] DelayedArray_0.30.1     abind_1.4-8            
[21] tidyselect_1.2.1        locfit_1.5-9.12        
[23] stringi_1.8.7           labeling_0.4.3         
[25] colorspace_2.1-1        cli_3.6.5              
[27] SparseArray_1.4.8       magrittr_2.0.3         
[29] S4Arrays_1.4.1          utf8_1.2.5             
[31] withr_3.0.2             scales_1.4.0           
[33] UCSC.utils_1.0.0        bit64_4.6.0-1          
[35] timechange_0.3.0        lambda.r_1.2.4         
[37] XVector_0.44.0          httr_1.4.7             
[39] bit_4.6.0               hms_1.1.3              
[41] rlang_1.1.6             futile.options_1.0.1   
[43] Rcpp_1.0.14             glue_1.8.0             
[45] BiocManager_1.30.25     formatR_1.14           
[47] vroom_1.6.5             rstudioapi_0.17.1      
[49] jsonlite_2.0.0          R6_2.6.1               
[51] zlibbioc_1.50.0
DESeq2 • 540 views
ADD COMMENT
0
Entering edit mode

I used

dds <- DESeq(dds)
resultsNames(dds)

[1] "Intercept"             "group_treatment1_vs_longer_disease"    
[3] "group_healthy_vs_longer_disease" "group_treatment2_vs_longer_disease"    
[5] "group_treatment2_vs_longer_disease"  "group_disease_vs_longer_disease"

unlike before, the treatments do not seem to have an effect... So I take it that LTR works to show genes that change across all my groups? The actual results of LRT make sense biologically, but I am concerned about the statistic robustness, so I can go for a padj < 0.01? Thanks again

ADD REPLY

Login before adding your answer.

Traffic: 1066 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6