RNA-STAR Failing using hg38 but works for hg19

Question: RNA-STAR Failing using hg38 but works for hg19

6 months ago by

aklie • 0

aklie • 0 wrote:

Hi,

I am trying to align two collections of 4 single-end RNA-Seq fastq files (8 fastq files total) from human skin samples to a reference genome using RNA STAR. I have no problem aligning to the hg19 reference genome but run into memory issues when aligning to the hg38. I've tried twice for each reference genome and have been successful twice with hg19 and failed twice with hg38. The error message I receive when using hg38 as the reference is:

"This job was terminated because it used more memory than it was allocated."

I'm hoping someone can help me understand what's going on here and what I might do to fix it!

Thanks!

star memory usage rna-seq hisat2 • 268 views

ADD COMMENT • link •

modified 6 months ago by Jennifer Hillman Jackson ♦ 25k • written 6 months ago by aklie • 0

6 months ago by

Jennifer Hillman Jackson ♦ 25k

United States

Jennifer Hillman Jackson ♦ 25k wrote:

Hello,

The job is probably right at the edge of the memory limits available on our clusters. I would suggest trying a few more test runs versus hg38. Nodes can process more that one job at a time, and by chance sometimes two large jobs end up on the same node, triggering this type of error.

You might also consider doing some more QA/QC on your input data, it might help. Reviewing tool options and testing out alternatives that are not as compute-intensive can also lead to successful jobs.

If that doesn't work out, it is good to know that RNA STAR uses a lot of memory. The target genome size is one important factor, although I know that many users map against hg38 successfully with STAR, so it is not the only factor.

When there are failures with STAR that cannot be resolved with input data/parameter changes, HISAT2 is the recommended alternative.

The settings for HISAT2 when matter and need to be adjusted for the downstream tools you plan to use. Please see the Galaxy RNA-seq tutorials here for example usage, including QA/QC: https://galaxyproject.org/learn/

HISAT2 spliced mapping is specifically covered here: https://galaxyproject.org/tutorials/nt_rnaseq/#spliced-mapping-with-hisat

Support FAQs: https://galaxyproject.org/support

For troubleshooting or to better understand errors (including memory failures), please see the FAQs here https://galaxyproject.org/support/#unexpected-results and here https://galaxyproject.org/support/#getting-inputs-right.

Thanks! Jen, Galaxy team

ADD COMMENT • link modified 6 months ago • written 6 months ago by Jennifer Hillman Jackson ♦ 25k

Thank you for such a detailed answer! That makes some sense given STAR memory requirements and the fact that the hg38 is a larger reference for STAR to build a suffix array. I wanted to run through pipelines with multiple different aligners anyway so I'll see if I can get it working and if not I just won't include it. Thanks for your help!

ADD REPLY • link written 6 months ago by aklie • 0

Similar posts • Search »