The freely available alignment program mzmine can be very fast. However to enjoy
speed, if you do not use the computer-cluster functionality, you have to adjust your starting batch file
startMZmine.bat with the following commands:
startMZmine.bat:
java -server -Xms1024m -Xmx1024m -XX:+AggressiveHeap -jar MZmine.jar
NumberOfNodes=4
This will start mzmine with the java server version (J2EE), which gives almost
20-30% speed improvement. You must check if you have in your java bin directory the subdirectories
client and server. If you have a dual core or dual processor machine you can use 2 nodes. If your hard
disk array (like an Areca-1120RAID5)
provides enough bandwidth, you can set the number of nodes to 4, even if you do
not have so many cores, however speed will improve a little bit due to faster threading (using more
nodes than real processing threads are available usually reduces speed with large files, with small
files it works). You should have at least 2 GB RAM or otherwise you have to adjust to -Xms512m
-Xmx512m.
A complete alignment procedure (mark all files and use mzmine "define batch mode" and set all options to
the first possible option; the result window should have 3 alignment files) with the faahKOsample
data runs in 15 seconds and theCroseus example in 3 minutes, using a Dual
Opteron 254 (2.8 GHz) with an ARECA-1120 RAID5. This is fun!
Picture Source:BMC Bioinformatics2005,6:179 (Open Access)