Hi, 

I am using RNA-Seq data for WGCNA. I have 34 samples. For my WGCNA analysis, I am using o networkType=Signed hybrid, TOM=Signed, corType=bicor, pearsonFallback = "individual" & deepSplit= 2 .

nethybrid.2 = blockwiseModules(datExpr, power = softpower,maxBlockSize = 46000,
                                     TOMType = "signed", minModuleSize = 30, deepSplit =2,
                                     reassignThreshold = 0, mergeCutHeight = 0.25,
                                     numericLabels = TRUE, pamRespectsDendro = FALSE,
                                     saveTOMs = TRUE,networkType = "signed hybrid",
                                     saveTOMFileBase = "34patient_signedhybrid_TOM_46000",
                                     verbose = 5,corType = "bicor", maxPOutliers = 0.1,
                                     pearsonFallback = "individual")

I know maximum WGCNA can analyze is 46000. That is why I decreased my genes to 45901. My aim is to analyse them all together in 1 block to get 1 TOM file for further network analysis. However, when I run the code below, my genes are divided into 2 block. Is there any possible way to prevent this division into multiple blocks ? Or is it possible if only I follow the steps in the step by step WGCNA tutorials. Because I also tried it, but it hangs everytime.

When  I include this line of command  saveTOMs = TRUE, saveTOMFileBase = "34patient_signedhybrid_TOM_46000"  to save TOM, it runs very long and never finishes. Is there anyway to optimize it ?  I read this post WGCNA blockwiseModules parallelisation question suggesting fast Blast library, but I though as my data is much more smaller than their data, there might be another solution also ?

Thanks in advance,

Gokce

> sessionInfo()
R version 3.3.1 (2016-06-21)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: CentOS release 6.5 (Final)

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
 [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

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

other attached packages:
 [1] EnsDb.Hsapiens.v79_0.99.12 ensembldb_1.4.7           
 [3] edgeR_3.14.0               limma_3.28.21             
 [5] amap_0.8-14                sva_3.20.0                
 [7] mgcv_1.8-15                nlme_3.1-128              
 [9] doParallel_1.0.10          iterators_1.0.8           
[11] foreach_1.4.3              reshape_0.8.5             
[13] cluster_2.0.4              matrixStats_0.50.2        
[15] flashClust_1.01-2          WGCNA_1.51                
[17] fastcluster_1.1.21         dynamicTreeCut_1.63-1     
[19] pheatmap_1.0.8             genefilter_1.54.2         
[21] gplots_3.0.1               RColorBrewer_1.1-2        
[23] vsn_3.40.0                 org.Hs.eg.db_3.3.0        
[25] DESeq2_1.12.4              BiocParallel_1.6.6        
[27] GenomicAlignments_1.8.4    SummarizedExperiment_1.2.3
[29] GenomicFeatures_1.24.5     AnnotationDbi_1.34.4      
[31] Biobase_2.32.0             Rsamtools_1.24.0          
[33] Biostrings_2.40.2          XVector_0.12.1            
[35] GenomicRanges_1.24.3       GenomeInfoDb_1.8.7        
[37] IRanges_2.6.1              S4Vectors_0.10.3          
[39] BiocGenerics_0.18.0        Hmisc_3.17-4              
[41] ggplot2_2.1.0              Formula_1.2-1             
[43] survival_2.39-5            lattice_0.20-34           

loaded via a namespace (and not attached):
 [1] httr_1.2.1                    AnnotationHub_2.4.2          
 [3] splines_3.3.1                 gtools_3.5.0                 
 [5] shiny_0.14                    interactiveDisplayBase_1.10.3
 [7] affy_1.50.0                   latticeExtra_0.6-28          
 [9] impute_1.46.0                 RSQLite_1.0.0                
[11] digest_0.6.10                 chron_2.3-47                 
[13] colorspace_1.2-6              httpuv_1.3.3                 
[15] htmltools_0.3.5               preprocessCore_1.34.0        
[17] Matrix_1.2-7.1                plyr_1.8.4                   
[19] XML_3.98-1.4                  biomaRt_2.28.0               
[21] zlibbioc_1.18.0               xtable_1.8-2                 
[23] GO.db_3.3.0                   scales_0.4.0                 
[25] gdata_2.17.0                  affyio_1.42.0                
[27] annotate_1.50.0               nnet_7.3-12                  
[29] mime_0.5                      foreign_0.8-67               
[31] BiocInstaller_1.22.3          tools_3.3.1                  
[33] data.table_1.9.6              munsell_0.4.3                
[35] locfit_1.5-9.1                caTools_1.17.1               
[37] grid_3.3.1                    RCurl_1.95-4.8               
[39] bitops_1.0-6                  gtable_0.2.0                 
[41] codetools_0.2-14              DBI_0.5-1                    
[43] R6_2.1.3                      gridExtra_2.2.1              
[45] rtracklayer_1.32.2            KernSmooth_2.23-15           
[47] Rcpp_0.12.7                   geneplotter_1.50.0           
[49] rpart_4.1-10                  acepack_1.3-3.3              

You need to set the argument 'maxBlockSize' for blockwiseModules to something larger than your number of genes. But do note that to analyze a 45k genes in a single block, you need a workstation with at least 64GB available RAM; 96GB or more may be preferable. If you have less RAM, the computer will start swapping to disk (using the disk as "supplemental" RAM) which is orders of magnitude slower and feels like the calculation never finishes.

Also note that saving a TOM that is around 10GB in file size will take a long time (depending on how fast the file system is).

Indeed, because of the way the compiled code is called, WGCNA cannot at present analyze more than sqrt(2^31) ~ 46300 genes in one block; but it can analyze more in the block-by-block manner.