Dealing with paired-end shotgun metagenomics sequencing ran on two lanes

I’m new to metagenomics and I have been using bioBakery’s pipeline to process shotgun metagenomics sequencing data that were ran on children stool samples.

I am in the process of doing the data cleaning, i.e. trimming of adaptor sequences and cleaning of contaminants. I have been running Kneaddata on my pair-end sequencing data to carry out the data cleaning. My question is what is the best way to deal with paired end data if they were ran on two lanes.

Should I run Kneaddata on the two paired-end sequencing data on different lanes separately first before concatenating? or concatenate them first before running kneaddata? Because after that running Humman3 required for concatenation of the paired data again.

Any advise will be appreciated!