Hello,
I'm running the Manananggal console application to identify alternative splicing events for 69 paired samples and am wondering if there's a way to speed up the run time. I have generated all the necessary input files (BigWig, count files, merged junction count files) and am invoking the Manananggal console application on a node with 32 cores and 251GB of RAM with the following command:
java -cp /lib/*:Manananggal_v1.0.7.jar Manananggal.SplicingAnalyzer gencode.v19.annotation.gtf CD4vsCD14.project condition 3 5 0.7 0.05 cross_ref_human.txt -skipFirstAndLast -skipIntronDetection -threads=20
Results are getting concatenated to the output files, so I know the application is running free of errors but it's running very slow. For one of my analyses it's taken 3.5 days to process ~10,000 of the 57,820 genes in the GENCODE v19 annotation file. This same analysis took less than 24 hours for DEXSeq (without gene aggregation).
Changing the maximum number of threads and using alternative parameter sets (5_5_0.7_0.1_true) does not make a significant difference in run time. Is there a way to either a) speed up the application or b) analyze subsets of genes so that I could analyze different subsets in parallel? I was very excited to discover and starting using Manananggal. I think the viewer is fantastic so I really hope to reduce the run time of the console application to something reasonable.
Thanks in advance.
Can you tweak xmx parameter and increase the memory available to jvm?
Thank you for your suggestion. Unfortunately, setting the xmx parameter didn't make any difference either.