Hi,

I would like to deeply understand the output of the function: I have trouble in the resulting table, when I retrieve data as data.frame

When I read the resulting table I cannot reproduce the fold change with the saved values, if I use the option bNormalized=TRUE I include an example:

Chr Start   End Conc    **Conc_CD1C_POS**   **Conc_NEG**    **Fold**    p-value FDR 

sample1 sample2 sample3 sample4 sample5 sample6 Called1 Called2 102138 chr2 137707013 137707413 8,234421435 1,649371384 9,22688822 -6,20142532 1,63485E-23 2,91429E-18 0 5,76 3,65 1,649232222 745,3 331,5 720,79 9,226885788 3 3

The question is: How could be interpreted the fold and reproduced using the values reported? The problem is not present if using bNormalized=FALSE. But the fold is calculated on normalized data, right? based on the analysis method used

thanks a lot,

Michela

 dba.report(DBA, contrast, method=DBA$config$AnalysisMethod, 
           th=DBA$config$th, bUsePval=DBA$config$bUsePval, 
           fold=0, bNormalized=TRUE, bFlip=FALSE, precision,
           bCalled=FALSE, bCounts=FALSE, bCalledDetail=FALSE,
           bDB, bNotDB, bAll=TRUE, bGain=FALSE, bLoss=FALSE,
           file,initString=DBA$config$reportInit,ext='csv',
           DataType=DBA$config$DataType) 



sessionInfo()
R version 4.1.1 (2021-08-10)
Platform: x86_64-apple-darwin17.0 (64-bit)
Running under: macOS Mojave 10.14.6

Matrix products: default
BLAS:   /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/4.1/Resources/lib/libRlapack.dylib

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

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

other attached packages:
 [1] openxlsx_4.2.4              DiffBind_3.2.6             
 [3] SummarizedExperiment_1.22.0 Biobase_2.52.0             
 [5] MatrixGenerics_1.4.3        matrixStats_0.60.1         
 [7] GenomicRanges_1.44.0        GenomeInfoDb_1.28.2        
 [9] IRanges_2.26.0              S4Vectors_0.30.0           
[11] BiocGenerics_0.38.0        

loaded via a namespace (and not attached):
  [1] backports_1.2.1          GOstats_2.58.0          
  [3] BiocFileCache_2.0.0      plyr_1.8.6              
  [5] GSEABase_1.54.0          splines_4.1.1           
  [7] BiocParallel_1.26.2      ggplot2_3.3.5           
  [9] amap_0.8-18              digest_0.6.27           
 [11] invgamma_1.1             GO.db_3.13.0            
 [13] SQUAREM_2021.1           fansi_0.5.0             
 [15] magrittr_2.0.1           checkmate_2.0.0         
 [17] memoise_2.0.0            BSgenome_1.60.0         
 [19] base64url_1.4            limma_3.48.3            
 [21] Biostrings_2.60.2        annotate_1.70.0         
 [23] systemPipeR_1.26.3       bdsmatrix_1.3-4         
 [25] prettyunits_1.1.1        jpeg_0.1-9              
 [27] colorspace_2.0-2         blob_1.2.2              
 [29] rappdirs_0.3.3           apeglm_1.14.0           
 [31] ggrepel_0.9.1            dplyr_1.0.7             
 [33] crayon_1.4.1             RCurl_1.98-1.4          
 [35] jsonlite_1.7.2           graph_1.70.0            
 [37] genefilter_1.74.0        brew_1.0-6              
 [39] survival_3.2-13          VariantAnnotation_1.38.0
 [41] glue_1.4.2               gtable_0.3.0            
 [43] zlibbioc_1.38.0          XVector_0.32.0          
 [45] DelayedArray_0.18.0      V8_3.4.2                
 [47] Rgraphviz_2.36.0         scales_1.1.1            
 [49] pheatmap_1.0.12          mvtnorm_1.1-2           
 [51] DBI_1.1.1                edgeR_3.34.0            
 [53] Rcpp_1.0.7               xtable_1.8-4            
 [55] progress_1.2.2           emdbook_1.3.12          
 [57] bit_4.0.4                rsvg_2.1.2              
 [59] AnnotationForge_1.34.0   truncnorm_1.0-8         
 [61] httr_1.4.2               gplots_3.1.1            
 [63] RColorBrewer_1.1-2       ellipsis_0.3.2          
 [65] pkgconfig_2.0.3          XML_3.99-0.7            
 [67] dbplyr_2.1.1             locfit_1.5-9.4          
 [69] utf8_1.2.2               tidyselect_1.1.1        
 [71] rlang_0.4.11             AnnotationDbi_1.54.1    
 [73] munsell_0.5.0            tools_4.1.1             
 [75] cachem_1.0.6             generics_0.1.0          
 [77] RSQLite_2.2.8            stringr_1.4.0           
 [79] fastmap_1.1.0            yaml_2.2.1              
 [81] bit64_4.0.5              zip_2.2.0               
 [83] caTools_1.18.2           purrr_0.3.4             
 [85] KEGGREST_1.32.0          RBGL_1.68.0             
 [87] xml2_1.3.2               biomaRt_2.48.3          
 [89] compiler_4.1.1           rstudioapi_0.13         
 [91] filelock_1.0.2           curl_4.3.2              
 [93] png_0.1-7                geneplotter_1.70.0      
 [95] tibble_3.1.4             stringi_1.7.4           
 [97] GenomicFeatures_1.44.2   lattice_0.20-44         
 [99] Matrix_1.3-4             vctrs_0.3.8             
[101] pillar_1.6.2             lifecycle_1.0.0         
[103] BiocManager_1.30.16      irlba_2.3.3             
[105] data.table_1.14.0        bitops_1.0-7            
[107] rtracklayer_1.52.1       R6_2.5.1                
[109] BiocIO_1.2.0             latticeExtra_0.6-29     
[111] hwriter_1.3.2            ShortRead_1.50.0        
[113] KernSmooth_2.23-20       MASS_7.3-54             
[115] gtools_3.9.2             assertthat_0.2.1        
[117] DESeq2_1.32.0            Category_2.58.0         
[119] rjson_0.2.20             withr_2.4.2             
[121] GenomicAlignments_1.28.0 batchtools_0.9.15       
[123] Rsamtools_2.8.0          GenomeInfoDbData_1.2.6  
[125] hms_1.1.0                grid_4.1.1              
[127] DOT_0.1                  coda_0.19-4             
[129] GreyListChIP_1.24.0      ashr_2.2-47             
[131] mixsqp_0.3-43            bbmle_1.0.24            
[133] numDeriv_2016.8-1.1      restfulr_0.0.13

This is a bit of a quirk really. As you say, the analysis methods compute the fold changes using normalized data, so it is not entirely clear what to report when raw data is requested.

When bNormalized=TRUE, the fold change is reported as computed by the analysis method. When bNormalized=FALSE, the fold change is computed by simply subtracting the log2 of the mean of the raw read counts across the samples in each group of the contrast. The confidence statistics (FDR) are always based on normalized data, though. I'll have a look at making this all more explicit in the help page for dba.report().

You can run the report both ways and then mix and match the metadata you are interested in (ie., the shrunken fold changes computed by DESeq2 combined with the raw read counts for each of the samples).