The amount of memory is hard-coded in the STAR-fusion wrapper
The problem
The memory usage for STAR is hard coded in the STAR-Fusion script.
The workaround
To increase the available memory for hard sorting. I had to manually update the --limitBAMsortRAM
option in /exports/hem-leukemia/aml-downsample/conda-run/062d838e/lib/STAR-Fusion/STAR-Fusion
wrapper script of the conda environment.
Permanent solution
Look into whether this has been fixed in a later version of STAR-Fusion