WGCNA error in network construction ?
0
0
Entering edit mode
m.azhar ▴ 10
@mazhar-12584
Last seen 6.8 years ago

Hi I am using WGCNA for network construction  but I am using tutorial script for automatic network construction

and i get this error i dont understand what this mean

1.> # Choose a set of soft-thresholding powers
> powers = c(c(1:10), seq(from = 12, to=20, by=2))
> # Call the network topology analysis function
> sft = pickSoftThreshold(datExpr, powerVector = powers, verbose = 5)
pickSoftThreshold: will use block size 720.
 pickSoftThreshold: calculating connectivity for given powers...
   ..working on genes 1 through 720 of 62103
Error in { : task 1 failed - "'x' has a zero dimension."
In addition: Warning message:
executing %dopar% sequentially: no parallel backend registered 
> # Plot the results:
> sizeGrWindow(9, 5)

 

 

microarray rnaseq • 5.7k views
ADD COMMENT
0
Entering edit mode

What is the dimension of your datExpr? Could you post the output of dim(datExpr)? Maybe there aren't any samples left after the quality control of goodSamplesGenes.

ADD REPLY
0
Entering edit mode

Thanks I used  dataExpr0 it has the data instead of Expr so it worked but when going to calculate TOM 

it gives me this error

 I am running this scrip in C directory and it has 60 Gb free space so why it still gives this error 

> coexpression<-adjacency(data)
Error: cannot allocate vector of size 28.7 Gb
In addition: Warning messages:
1: In cor(datExpr, use = "p") :
  Reached total allocation of 3810Mb: see help(memory.size)
2: In cor(datExpr, use = "p

 

but I have spc

 

ADD REPLY
0
Entering edit mode

Sorry I don't understand what do you mean when you say "I used  dataExpr0 it has the data instead of Expr " could you post a piece of your data? If you already calculated the TOM why are you later calculating the adjacency?

What do you mean by "I have spc"?

ADD REPLY
0
Entering edit mode

I mean i used tutrioal object Expr but it has no data and i chekced bu str(Expr0) object Expr0 has the data so i used Expr0 object further in the script and data look  like 

> str(Expr)
Error in str(Expr) : object 'Expr' not found
> str(datExpr0)
'data.frame':    18 obs. of  62103 variables:
 $ : num  9.19 9.57 9.39 9.54 9.11 ...
 $ : num  10.97 11.3 10.13 8.69 10.3 ...
 $ : num  8

and I have not calculated the TOM i am going to calculate Adjacency matrix then i will calculat the TOM ?

 

some body sugessted me that the above error mentioned in the question is not hard disk space error but its less RAM error is it right ?  

ADD REPLY
0
Entering edit mode

Dear Muhammad,

there is no object Expr, you need to use datExpr. It seems that you are new to R and possibly to larger-scale computation and you may benefit from learning more about R and finding someone local who can guide you through some of the pitfalls. You seem to be using an R installation which can only use 4GB of memory. Please read the WGCNA tutorial I, section 2c (analysis of large data) carefully since your computer RAM is too small to handle the large data set (plus, WGCNA currently cannot handle more than about 46300 variables in one block). You may also want to look at WGCNA FAQ at https://labs.genetics.ucla.edu/horvath/CoexpressionNetwork/Rpackages/WGCNA/faq.html regarding gene filtering. There is also a short article on analysis of large data in WGCNA at http://www.peterlangfelder.com/blockwise-network-analysis-of-large-data/ .

ADD REPLY
0
Entering edit mode

Dear Peter 

                     Thanks for advice yes you are right actually I am learning how to work in r. I used short code 

like 

data<-t(data)
coexpression<-adjacency(data) 

using  remote server now i want to convert save this file into .csv convert into cytoscape or visant file network file

ADD REPLY
0
Entering edit mode

yes you are right i read through these links before and again but i did not found any solution because I want to see lnc RNA and mRNA correlation network and one file contains more then 60 k probes i used block wise approach as suggested by WGCNA for 2000 but it did not work so what should i do guide me any solution

 

ADD REPLY
0
Entering edit mode

What do you mean by "it did not work", which error/problem did you have?

ADD REPLY
0
Entering edit mode

I mean blockwise=2000 argument did not work , I used a server with 64 GB ram and have made network file but when I want to export the file I have to calculate The TOM according to WGCNA tutrioal but it TOM for data object gives no error but after some time it just do not go forward like 

> TOM=TOMsimilarityFromExpr(datExpr0, power =6);
TOM calculation: adjacency..
adjacency: replaceMissing: 0
..will use 23 parallel threads.
 Fraction of slow calculations: 0.000000
..connectivity..
..matrix multiplication

whats the problem

can i make export file from network.Radat file directly 

 

 

 

ADD REPLY
0
Entering edit mode

It is normal to spend some time in that step. of "...matrix multiplication". Be patient

ADD REPLY
0
Entering edit mode

how much time because yesterday I tried 4 times and after one hour at this step connection to server was disconnected, can I ask is this step is necessary to get export file or we can skip this 

ADD REPLY
0
Entering edit mode

The time spend depends on the server. You can use nohup to keep the process going even if you disconnect from the server (I am assuming a Unix server). You can use blockwiseModules to do it automatically, see tutorial 1) 2.c

ADD REPLY
0
Entering edit mode

yes you right and thanks for input and advice, I already used block wise module approach for for network.Rdata file but for TOM=TOMsimilarityFromExpr(datExpr0, power =6); can i use bolck wise argument again

?

or is there some other way to use network.Rdata file form blockwise module detection to cytoscape or vis net file 

ADD REPLY
0
Entering edit mode

There is a function in WGCNA to export the network too cytoscape and visant, look in the manual. You only need the adjacency matrix.

ADD REPLY
0
Entering edit mode

yes that is final step names as  Exporting network data to network visualization software """ Exporting to VisANT The package provides a convenient function for exporting the network to VisANT [1]. We illustrate a simple export of the full weighted network of a single module. # Recalculate topological overlap TOM = TOMsimilarityFromExpr(datExpr, power = 6);"""""

so here I am stuck my question is when WGCNA has calculate adjcency for modue detection in block wise approach also TOMs

bwnet = blockwiseModules(datExpr, maxBlockSize = 2000, power = 6, TOMType = "unsigned", minModuleSize = 30, reassignThreshold = 0, mergeCutHeight = 0.25, numericLabels = TRUE, saveTOMs = TRUE, saveTOMFileBase = "femaleMouseTOM-blockwise", verbose = 3) 

after this step is it nessary to calcaulte TOM again like TOM=TOMsimilarityFromExpr(datExpr0, power =6);

ADD REPLY
0
Entering edit mode

I think you should use datExpr instead of datExpr0. But check with your advisor.

ADD REPLY
0
Entering edit mode

Thanks A lot for help Lluis but my Supervisor is not bioinformatic  so I have no help any I have move forward I used ssh 32 bit server to connect the linux based sever so I got TOM after 3.5 hours but i have two errors

dim(TOM) 62103 62103 

1. i wanted to save this data object but was not able to save i dont know why may be matrix is two large it gives error

error in writable connection 

2. when i wanted to save TOM as VisANT networl file 

 vis=exportNetworkToVisANT(modTOM, file=paste( "VisANTInput-", module, ".txt", Warning messages:TRUE, threshold=0)
1: In max(abs(adjMat - t(adjMat)), na.rm = TRUE) :
  no non-missing arguments to max; returning -Inf
2: In min(adjMat, na.rm = TRUE) :
  no non-missing arguments to min; returning Inf
3: In max(adjMat, na.rm = TRUE) :
  no non-missing arguments to max; returning -Inf
4: In matrix(c(1:nRow), nRow, nRow, byrow = TRUE) :
  data length exceeds size of matrix
5: In matrix(c(1:nRow), nRow, nRow) : data length exceeds size of matrix

what is the solution 

ADD REPLY
0
Entering edit mode

I am sorry to hear your problems:

1) the error seems to indicate you, or the process, got disconnected from the server. Or that it is too big to be handled properly.

2) That are warnings, not errors so you have already the network in visant format or whatever exportNetworkToVisANT returns in the vis variable.

Please, next time try to find yourself the solutions and explain what you found, it is a good skill to know how to solve problems. Also, I won't be able to always solve your problems :D

ADD REPLY

Login before adding your answer.

Traffic: 942 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