Running DIAMOND and MEGAN on very large files of reads can be challenging, e.g. when faced with time restrictions for a single job on a server.
To address this, divide your input file into multiple parts. Run each part separately through DIAMOND and meganizer.
Then use the new MEGAN tool merge-files to logically merge the set of meganized .DAA files into a single .MEGAN file. You can work with this file in MEGAN as if it contained all the input reads and their alignments. Under the hood, MEGAN iterates over all the input .DAA files when it needs to access the underlying data.
This is a new feature, available in release 6.23.