Holiday Notice:
The Frontline Support team will be slow to respond December 17-18 due to an institute-wide retreat and offline December 22- January 1, while the institute is closed. Thank you for your patience during these next few weeks. Happy Holidays!

PrintReads keeps running out of memory (GC overhead limit exceeded)

I am using base recalibration on human whole genome data and I keep running out of memory. I am running this on a cluster with plenty of memory but it seems that as I increase memory PrintReads is still using up what I provide to it. After trying 1 and 2 cores with 16GB in the java execution (-Xmx) line, in the most recent iteration, I requested 5 cores, each with 20GB of memory and executed the following command:

java -Xmx18g -jar GenomeAnalysisTK-3.4-46.jar -T PrintReads -nct 5 -R /gatk-resource/human_g1k_v37_decoy.fasta -I 2-18-1.BWA_aln_umi.rmdupumi.mdup.sorted.realigned.bam -BQSR 2-18-1.BWA_aln_umi.rmdupumi.mdup.sorted.realigned.recal.table -o 2-18-1.BWA_aln_umi.rmdupumi.mdup.sorted.realigned.recalibrated.bam

As before, I once again received the error:
java.lang.OutOfMemoryError: GC overhead limit exceeded

The cluster scheduler (LSF) reports:
Max Memory: 19267 MB
so clearly the 20GB requested is getting used.

For what it's worth, the program output seems to show that PrintReads is getting stuck on a relatively innocuous region of chromosome 11 that has 40X coverage. I can "samtools view" the suspect coordinate (+/- 100 bases without issue).

INFO 13:13:42,792 ProgressMeter - 11:98220738 3.81160926E8 5.9 h 55.0 s 61.0% 9.6 h 3.8 h
INFO 13:14:12,793 ProgressMeter - 11:100977123 3.81760933E8 5.9 h 55.0 s 61.1% 9.6 h 3.7 h
INFO 13:14:50,040 ProgressMeter - 11:101826859 3.81860934E8 5.9 h 55.0 s 61.1% 9.6 h 3.8 h
INFO 13:15:31,039 ProgressMeter - 11:101826859 3.81860934E8 5.9 h 55.0 s 61.1% 9.7 h 3.8 h
INFO 13:16:34,404 ProgressMeter - 11:101826859 3.81860934E8 5.9 h 55.0 s 61.1% 9.7 h 3.8 h
INFO 13:17:07,201 ProgressMeter - 11:101826859 3.81860934E8 5.9 h 55.0 s 61.1% 9.7 h 3.8 h
INFO 13:17:39,947 ProgressMeter - 11:101826859 3.81860934E8 5.9 h 56.0 s 61.1% 9.7 h 3.8 h
INFO 13:18:39,404 ProgressMeter - 11:101826859 3.81860934E8 6.0 h 56.0 s 61.1% 9.8 h 3.8 h
INFO 13:19:42,125 ProgressMeter - 11:101826859 3.81860934E8 6.0 h 56.0 s 61.1% 9.8 h 3.8 h
INFO 13:20:15,785 ProgressMeter - 11:101826859 3.81860934E8 6.0 h 56.0 s 61.1% 9.8 h 3.8 h
INFO 13:22:12,643 ProgressMeter - 11:101826859 3.81860934E8 6.0 h 56.0 s 61.1% 9.8 h 3.8 h
INFO 13:23:15,204 ProgressMeter - 11:101826859 3.81860934E8 6.0 h 56.0 s 61.1% 9.9 h 3.8 h
[then it dies]

I am trying to switch back to single thread and request 80GB of memory on a single node. Any other ideas?

Best Answer

Answers

Sign In or Register to comment.