search

sqjin / scAI

An unsupervised approach for the integrative analysis of single-cell multi-omics data
GNU General Public License v3.0
29 stars 8 forks source link

Task 1 failed #10

Closed Zhen-Miao closed 4 years ago

Zhen-Miao commented 4 years ago

Hi, I am running the program, but there seems to be an issue... Could you explain why is this?

Error in { : task 1 failed - "std::bad_alloc"

Thanks!

sqjin commented 4 years ago

Hi @Zhen-Miao , could you provide more information? I have no idea where this issue might happens.

Zhen-Miao commented 4 years ago

Hi @sqjin, thank you. I am running the program with R version 4.0.2 on the cluster. I was using 3000 top variable genes with 5000 variable peaks. Following your tutorial, I have built the scAI object scAI_out. Then I run the main function

scAI_outs <- run_scAI(scAI_outs, K = 20, nrun = 1)

Then it did not respond in 3 hours. I guess this was because I did not configure well the python packages, so I set do.fast = F, when this error occurred.

Please let me know if additional information is needed... Thanks!

Z

image

sqjin commented 4 years ago

@Zhen-Miao I just google such issue and it seems this is not a special issue given by scAI. Please see a similar issue here. I think this is related with the parallel code. We use the following way for parallel: # Calculate the number of cores numCores <- min(parallel::detectCores(), nrun) cl <- tryCatch({ parallel::makeForkCluster(numCores) }, error = function(e) { parallel::makeCluster(numCores) }) doParallel::registerDoParallel(cl) Therefore, you can test whether such parallel works on your cluster. You can also run your code on your local computer to see whether this issue happens again.

Zhen-Miao commented 4 years ago

@sqjin Thank you very much! I will check how this parallel computing can be achieved in my cluster.