In miRDeep you can create a config file containing multiple samples.

For instance you can make a config.txt containing the following lines (filename and 3 letter code):

wt1.fastq WT1
wt2.fastq WT2
wt3.fastq WT3
ko1.fastq KO1
ko2.fastq KO2

and you could call mapper.pl as follows (example):

mapper.pl config.txt --d --e --h --i --j --l 17 --m --p genome.index --s reads.fa --t reads vs genome.arf -v --o 4 --q

This is the "official" approach to this and this is definitely the way to go for quantification and for miRNA DE.

You could also make a config.txt only with the WTs and another with the KOs. This depends on your study design actually. I'll paste below the relevant passage from the online help.

Cheers,
IV

PS the relevant text from mirDeep2 doc

The user has sequencing data from different samples e.g. different cell-types. A config.txt file has to be created in which each line designates file locations and a unique 3 letter code. For instance:

sequencing_data_sample1.fa  sd1
sequencing_data_sample2.fa  sd2
sequencing_data_sample3.fa  sd3

The user wishes then to pool these files and use the generated files reads.fa and reads_vs_genome.fa for the miRDeep2 analysis.

mapper.pl config.txt -d -c -i -j -l 18 -m -p genome_index -s reads.fa -t reads_vs_genome.arf

Since the reads_vs_genome.arf still contains the 3 letter code for each read mapped to genome the user can then later on dilute the contribution of the different samples to a predicted or known miRNA. It can also be used for example to define 'high confident' predictions if the results are filtered for miRNAs that have sequencing evidence from at least two samples.