Memory Problems Running Gatk Printreads

Question

[ajc8]

Hi everyone,

I have been unable to run GATK Printreads as a part of my normal pipeline - I keep getting an error message stating that the startup disk has run out of memory. There is plenty of room on my Mac hard drive (~250 Gig), and I tried specifying a temp directory in the java command line, but that doesn't seem to help.

I have never had this problem before - I am doing one thing differently - running bwa mem instead of bwa aln and sampe. The raw data is illumina paired-end, exome. The fastq files are 7.3 GB each, which isn't any larger than files I have run in the past.

I am running the Printreads command after the BaseRecalibrator. And I'm specifying 8G RAM in the java command. My computer is a Macbook Pro i7 with 16G RAM. I haven't been using multiple cores.

Does bwa mem introduce a bug that isn't compatable with GATK Printreads?

Thanks for your help. Allison

Hi - this is the command line: java -Xmx8g -Djava.io.tmpdir=/Volumes/Passport/temp -jar GenomeAnalysisTK.jar -T PrintReads -R /Volumes/Passport/ref_genome/human_g1k_v37.fasta -I /Volumes/Passport/epilepsy_data/family2/Epi_362.realigned.fixed.dedup.bam -BQSR /Volumes/Passport/epilepsy_data/family2/Epi_362.recal_data.grp -o /Volumes/Passport/epilepsy_data/family2/Epi_362.realigned.fixed.dedup.recal.bam

The error message is from my Mac, saying there is no memory remaining to run any applications and then I have to force quit everything - I'm not able to copy and paste the message.

[Brad Chapman]

Allison -- could you include the command line and the error message? That would make it easier to offer suggestions about the potential problem.

[Brad Chapman]

Thanks for posting the command line. It's a good idea to post this as an edit to your original post, instead of as an answer to the question. Biostars is a bit different from standard forum sites since it separates question and answers; your information isn't an answer but rather an addition to the initial question. In terms of the actual problem, two suggestions:

• Try also setting the initial memory size (-Xms4g) in addition to your -Xmx8g option to java. Without this, programs can sometimes eat up lots of stack and cause memory issues.
• Are other processes running using significant memory? Using top in a console should let you monitor memory usage and identify what is using up your memory.

Hope this helps

[ajc8]

Thank you - it is running right now - we'll see what happens. I ran top and it says that java is using ~95% CPU and terminal is using around 85%. I don't usually see terminal taking up that much memory. Does that ring a bell?

and top is using up ~17% which is much higher than usual (~3%).

[Brad Chapman]

You should focus on monitoring memory usage since that's related to the error you're seeing. CPU usage won't help much with diagnosing the problem.

[Josh Herr]

Hi Allison, I just moved your error message post to your original question because this showed the question as "answered" which might cause people to think your problem was solved.

[ajc8]

thank you! I'm new at this

Answer 1

[ajc8]

Hi everyone, Thanks for all of your help.
It turns out it was an '\n' before the -o in my command that caused the problem - it must have snuck into the script somehow. So the PrintReads was printing everything to screen and overloading my computer. It is running just fine now. Allison