HaplotypeCaller Memory Optimization

HaplotypeCaller Memory Optimization


When using HaplotypeCaller on GATK, is there a fixed amount of memory that works well for for the java -Xmx input, or does it scale with the size of the input bam? eg if I have a 50 GB file do I need to set -Xmx to at least 50 GB to run haplotypecaller effectively, or can I use less memory than the file? The more general question would be “does the java heapsize need to be as large as an input file for a program run, or run efficiently?”

From this post, I could see that fewer threads was more efficient usage for HaplotypeCaller, and that less memory was better, but I wasn’t sure if that ideal memory allocation was relative to the size of their input .bam







Read more here: Source link