I have run kneaddata with paired-end data but i am getting some strange output. I have a forward and a reverse read file each of size 780MB. But when I run kneaddata it gives two output files of size 1.3 GB and 148MB. One file increased in size drastically and another decreased drastically.
I have checked the read counts of two input and catenated output file. I have seen that total read counts in the catenated output file is only few thousand less (among millions) than total read counts from forward and reverse files.
So, should I depend on the data and catenate the outputs followed by HUMAnN analysis? or, do not rely into the outputs?