Here are the first few lines of the help displayed by the program:
Computes a MEGAN .rma6 file from a DIAMOND .daa file
-i, --in [string(s)] Input DAA file. Mandatory option.
-mdf, --metaDataFile [string(s)] Files containing metadata to be included in RMA6 files.
-o, --out [string(s)] Output file(s), one for each input file, or a directory. Mandatory option.
-c, --useCompression Compress reads and matches in RMA file (smaller files, longer to generate. Default value: true.
-mag, --magnitudes Reads are annotated with magnitudes. Default value: false.
-p, --paired Reads are paired. Default value: false.
-ps, --pairedSuffixLength [number] Length of name suffix used to distinguish between name (i.e. first word in header) of read and its mate (use 0 if read and mate have same name). Default value: 0.
-pof, --pairedReadsInOneFile Are paired reads in one file (usually they are in two). Default value: false.
Use -p to tell the program that reads are paired, use -ps to provide the length of the suffix which is used to distinguish paired reads (e.g. use 1 if pairs are named XXX.1 and XXX.2).
Use -pof to tell the program that paired reads appear together in one file (otherwise you must provide two files).
The paired read mode is overly conservative at the moment, pairs of reads are assigned to the LCA of the LCA of each. I will implement a second strategy (based on adding bitscores) in the near future.