Hi,
Often, I see that authors compare different metagenomes from MG-RAST and finally conclude that "Pathway X" is stronger represented in the microbial community A than in community B.
However, in my opinion, this is a problem for host-associated metagenomes, since these contain also reads from the host organism.
If in Metagenome A contains 20% of host reads and metagenome B contains only 5% of host reads, basically all bacterial pathways would appear to be enriched in Metagenome B.
I have this concrete problem, because I want to compare the metagenome of several closely related host-species. But I want to compare only bacterial genes. Host reads range from 50% - 5% in different species (probably correlates with bacterial load in the gut).
How can I limit my metagenome analysis to bacteria only.
My thought was to blast my raw reads against all NCBI bacterial genomes and proceed only with reads that had a hit, but this is too computational demanding?
How to people do this. Seems like most just ignore the host-read fraction (since for luminal content/feces it is small anyway).
thanks.
Often, I see that authors compare different metagenomes from MG-RAST and finally conclude that "Pathway X" is stronger represented in the microbial community A than in community B.
However, in my opinion, this is a problem for host-associated metagenomes, since these contain also reads from the host organism.
If in Metagenome A contains 20% of host reads and metagenome B contains only 5% of host reads, basically all bacterial pathways would appear to be enriched in Metagenome B.
I have this concrete problem, because I want to compare the metagenome of several closely related host-species. But I want to compare only bacterial genes. Host reads range from 50% - 5% in different species (probably correlates with bacterial load in the gut).
How can I limit my metagenome analysis to bacteria only.
My thought was to blast my raw reads against all NCBI bacterial genomes and proceed only with reads that had a hit, but this is too computational demanding?
How to people do this. Seems like most just ignore the host-read fraction (since for luminal content/feces it is small anyway).
thanks.