I am slightly confused about how I go about binning my individually assembled metagenome samples. Below I lay out my current plan:

1) Run assemblies on each individual sample using megahit. This will give me multiple metagenome assemblies.

Confusion begins:

2.a) Merge all the contigs from different assemblies into a single contig file and use that for binning.

OR

2.b) Keep each assembly separate, and bin them separately? If I do this, how do I get a single file with all the bins?

3) Map reads back to bins to quantify abundance. If I do 2.b, then I will have different reference for each sample. This will make it hard to compare between samples.