Question: Different results obtained from mroast and camera from Deseq2 counts
0
24 months ago by
Italy
jarod_v6@libero.it40 wrote:

I have import data on deseq2 using tximport from rsem results.

After create the rlog object I  try to use mroast and camera on  some pathway. This is the script I use:

files <- list.files(path=path, pattern="*.ensembl.csv", full.names=T, recursive=FALSE)
tipo<-list()
result_camera<-data.frame("Ngenes"=numeric(),"Direction"=character(),"PValue"=numeric(),"name"=character(),stringsAsFactors = F)

result_mroast<-data.frame("NGenes" =numeric(),"PropDown"=numeric(),"PropUp"=numeric(),"Direction"=character(),"PValue"=numeric(),
"FDR"=numeric(),"PValue.Mixed"=numeric(), "FDR.Mixed" =numeric(),stringsAsFactors = F)
for(i in 1:length(files)){
ao<-strsplit(files[i],"/")[[1]]
name<-ao[9]
id=ids2indices(listaGeni$V1,rownames(v)) condition<-factor(metadata$CONDITION)

design<-model.matrix(formula(~ condition))
gst.camera<-camera(v,id,design = design)

pi<-mroast(v,index=id,design,nrot = 99)
result_mroast<-rbind(result_mroast,data.frame(pi))
result_camera<-rbind(result_camera,data.frame(gst.camera$NGenes,gst.camera$Direction,gst.camera$PValue,name)) }  There is some difference For camera for istance  28 Up 0.074871 Bcellsmemories.csv.ensembl.csv for the mroast: NGenes PropDown PropUp Direction PValue FDR PValue.Mixed FDR.Mixed Set1 28 0.00000000 0.25000000 Up 0.34 0.34 0.17 0.17  sessionInfo() R version 3.3.2 (2016-10-31) Platform: x86_64-pc-linux-gnu (64-bit) Running under: Ubuntu 16.04.2 LTS locale: [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C LC_TIME=it_IT.UTF-8 [4] LC_COLLATE=en_US.UTF-8 LC_MONETARY=it_IT.UTF-8 LC_MESSAGES=en_US.UTF-8 [7] LC_PAPER=it_IT.UTF-8 LC_NAME=C LC_ADDRESS=C [10] LC_TELEPHONE=C LC_MEASUREMENT=it_IT.UTF-8 LC_IDENTIFICATION=C attached base packages: [1] parallel stats4 stats graphics grDevices utils datasets methods [9] base other attached packages: [1] GSEABase_1.36.0 graph_1.52.0 annotate_1.52.1 [4] XML_3.98-1.9 AnnotationDbi_1.36.2 limma_3.30.13 [7] clusterProfiler_3.2.14 DOSE_3.0.10 biomaRt_2.30.0 [10] pheatmap_1.0.8 tximportData_1.2.0 tximport_1.2.0 [13] DESeq2_1.14.1 SummarizedExperiment_1.4.0 Biobase_2.34.0 [16] GenomicRanges_1.26.4 GenomeInfoDb_1.10.3 IRanges_2.8.2 [19] S4Vectors_0.12.2 BiocGenerics_0.20.0 loaded via a namespace (and not attached): [1] tidyr_0.6.3 bit64_0.9-7 splines_3.3.2 Formula_1.2-1 [5] DO.db_2.9 latticeExtra_0.6-28 blob_1.1.0 RSQLite_2.0 [9] backports_1.1.0 lattice_0.20-35 digest_0.6.12 RColorBrewer_1.1-2 [13] XVector_0.14.1 checkmate_1.8.2 qvalue_2.6.0 colorspace_1.3-2 [17] htmltools_0.3.6 Matrix_1.2-10 plyr_1.8.4 pkgconfig_2.0.1 [21] genefilter_1.56.0 zlibbioc_1.20.0 xtable_1.8-2 GO.db_3.4.0 [25] scales_0.4.1 BiocParallel_1.8.2 htmlTable_1.9 tibble_1.3.3 [29] ggplot2_2.2.1 nnet_7.3-12 lazyeval_0.2.0 survival_2.41-3 [33] magrittr_1.5 memoise_1.1.0 foreign_0.8-69 tools_3.3.2 [37] data.table_1.10.4 stringr_1.2.0 munsell_0.4.3 locfit_1.5-9.1 [41] cluster_2.0.6 rlang_0.1.1 grid_3.3.2 RCurl_1.95-4.8 [45] htmlwidgets_0.8 igraph_1.0.1 bitops_1.0-6 base64enc_0.1-3 [49] gtable_0.2.0 DBI_0.7 reshape2_1.4.2 gridExtra_2.2.1 [53] knitr_1.16 bit_1.1-12 fastmatch_1.1-0 Hmisc_4.0-3 [57] fgsea_1.0.2 stringi_1.1.5 GOSemSim_2.0.4 Rcpp_0.12.11 [61] geneplotter_1.52.0 rpart_4.1-11 acepack_1.4.1  limma camera mroast tximport • 601 views ADD COMMENTlink modified 24 months ago by James W. MacDonald50k • written 24 months ago by jarod_v6@libero.it40 Note that you can get the estimated gene counts straight from the tximport list: txi$counts, you don't need DESeq2 here.

so if I use tximport\$counts I obtain total different results...(if mroast see up as direction then I found down)

I don't see what the problem is. Why would you expect roast() and camera() to give the same results? They're different tests, so naturally they will give different results. In this case, neither test gives a statistically significant result, so they pretty much agree.

As I say, what's the problem?

I have this problem. Camera seem gave one results the mroast gave me  a little different results. You say this is normal. So the problem is what is the right method to choose when you question is to say if there is some enrichment in this list of data? with mroast don't have any significant change with camera yes

http://imgur.com/a/6dNqA

thanks for patience help

Answer: Different results obtained from mroast and camera from Deseq2 counts
0
24 months ago by
United States
James W. MacDonald50k wrote:

As Gordon already pointed out, mroast and camera do completely different tests! The right method to choose is the one that answers the biological question you are asking, not the one that gives you a significant result. If you are confused as to the difference, you should read the Goeman and Buhlmann paper that is referenced in both of the help files. Or you could simply read the help files for each of the functions you are using. You will find that you get better help if people think you are at least attempting to understand things for yourself, rather than expecting people on this forum to explain things that you could easily find for yourself. From ?mroast:

     'roast' and 'mroast' test whether any of the genes in the set are
differentially expressed. They can be used for any microarray
experiment which can be represented by a linear model. The design
matrix for the experiment is specified as for the 'lmFit'
function, and the contrast of interest is specified as for the
'contrasts.fit' function. This allows users to focus on
differential expression for any coefficient or contrast in a
linear model. If 'contrast' is not specified, then the last
coefficient in the linear model will be tested.

and from ?camera:

     'camera' and 'interGeneCorrelation' implement methods proposed by
Wu and Smyth (2012). 'camera' performs a _competitive_ test in the
sense defined by Goeman and Buhlmann (2007). It tests whether the
genes in the set are highly ranked in terms of differential
expression relative to genes not in the set. It has similar aims
to 'geneSetTest' but accounts for inter-gene correlation. See
'roast' for an analogous _self-contained_ gene set test.