I don’t understand the problem. In my understanding for example there are species “A” and species “B”. And for example pangenome of A=300 Mbp, and pangenome B = 200 Mbp. It means that if we have equal amount of “A” and “B” in sample, the mapping proporiton “A/B” will be ~ 3/2 ( independently from genome size of “A” and “B” ) so, to get real relative counts - we need to divide bowtie hits “A” on 3 and bowtie hits “B” on 2.
As I mentioned early - I’m afraid of not high-pricision results of Metaphlan since of searching on only marker genes (which are very small part of whole genome )
Also I provided some initial tests on Metaphlan and seems to got not very good results - I wrote about this in topic Metaphlan target microbiomes