In some cases, repeated mappings against the same reference will result in a dramatically reduced runtime because the internal data structure used for mapping the reads, which is reference specific, can be reused. This has been enabled by storing files in the system tmp folder as a caching mechanism. Only a certain amount of disk space will be used and once reaching the limit, the oldest files are cleaned up. Consequently, the reference data structure files will automatically have to be recreated if the cache was filled or the tmp folder was cleaned up.
The default space limit is 16 GB.
The reference cache size can be changed by creating a file called
readmapper.properties, with an entry
referencecachesize = <size in bytes>, for example
referencecachesize = 8589934592
A copy of this file should then be placed into the
settings folder in the following locations:
- Under the installation area of the CLC Server on the single server or master node.
- Under the installation area of the CLC Server of each job node in a job node setup.
- Under each grid worker folder defined in your grid presets on a grid node setup.