jennasimit / mgflashfm Goto Github PK

Hi Dr. Asimit,

I am working on applying MGfm to simulated datasets, which is single trait with two ancestries (groups) scenario. However, basically, I got completely empty results from MGfm on all datasets. I assumed I might wrongfully use MGfm, but I do want to know how to get the PIPs of all the testing SNPs from MGfm, so that I could analyze the results to find out where I got it wrong. For now, MGfm only provided the information/results regarding the credible sets, which are empty.

Thank you in advance. Your help would be greatly appreciated.

Best,
Zikun

Hi Dr. Asimit,

I tried to use MGflashfm on my local Mac, which worked well. My colleague and I found the reason that it didn't work on our cluster is due to "OpenJDK version 1.8.0" used on the cluster. Further, we found that Java is called by the package R2BGLiMS. The R2BGLiMS::JAM function has a parameter (extra.java.arguments) to pass the java arguments ([https://github.com/pjnewcombe/R2BGLiMS/blob/master/R/JAM.R].

When MGflashfm calls the R2BGLiMS::JAM function, it does not include the parameter "extra.java.arguments", as shown in the function calling in [https://github.com/jennasimit/MGflashfm/blob/master/R/flashfm-updated.R], line 23.
jam.results <- R2BGLiMS::JAM(marginal.betas = BETA, trait.variance = Vy,
cor.ref = refG, mafs.ref = mafs.ref, model.space.priors = list(a = 1,
b = length(BETA), Variables = names(BETA)), max.model.dim = maxcv,
n = N, xtx.ridge.term = 0.01, save.path = save.path, n.mil.iter=jam.nM.iter)

The parameter "extra.java.arguments" is omitted in MGflashfm. To increase the java memory allocation for R2BGLiMS::JAM function, the parameter "extra.java.arguments" is needed to be added in "flashfm-updated.R".

It would be greatly appreciated if the package could be updated to address this parameter. Such an update would significantly enhance the package's compatibility with various operating systems. Thank you for considering this request.

Best,
Zikun

Hi Dr. Asimit,

I encountered another issue here. If I switched the option of "maxcv_autocheck" to "F" or "FALSE", the function "JAMdynamic" or "MGFMwithJAM" would return the error message "object 'JAM_output' not found". I have loaded both packages "MGflashfm" and "R2BGLiMS". It seems the function would work only with "maxcv_autocheck" setting to "TRUE". Could you let me know if I did something wrong here? Thank you for your time.

Best,
Zikun

hello, i hope my message finds you well!

i am interested in using mgflashfm for multi group, multi trait analysis (2 groups, 2 traits). however, i'm running into some issues that i'm not sure how to fix.

would you have any advice? thank you in advance

Hello!

Thanks for making this software. I'm trying to use MGfm for single-trait, multi-ancestry fine-mapping.

I'm having a hard time getting the example to work (also having a hard time with my data, but I figured I would start with the example). When I run the following code (taken from the demonstration on the website):

require(MGflashfm)
T1gwas.list <- list(gwas.list[[1]][[1]], gwas.list[[2]][[1]])
N1all <- sapply(Nall,function(x) x[1])
save.path <- '/n/scratch3/users/j/jor6523/mgfm'
mgCS1 <- MGFMwithJAM(T1gwas.list, corX.list, N1all, save.path,cpp=0.99,cred=0.99, maxcv=1, maxcv_stop = 20, maxcv_autocheck = TRUE,NCORES=1,jam.nM.iter=1)

I get the following error message:

Error in JAMcor.tries.maxcv(beta.list[[i]], corX.list[[i]], raf.list[[i]], :
object 'JAM_output' not found
Calls: MGFMwithJAM -> -> lapply -> FUN -> JAMcor.tries.maxcv
Execution halted

Do you have any idea why the example might not be working for me? I installed the package today using a fresh R 4.3.1 on a CentOS Linux machine. I've attached a file with the full output in case that's of any help.
err.txt

Thanks!
Jordan

hello! sorry to bother you again! i'm running mgflashfm on a different dataset and i seem to be getting another error:

any advice would be really appreciated!

thanks!