Can I use atlas even If I don't get good MAGs out of it?

[SilasK]

Question from a user:

I was looking through the Atlas metagenome tutorial and I had a question. I'm interested in quantifying the functional content (e.g., richness of Kegg orthologs, I suppose) in a set of metagenomes. However, the metagenomes that I'm working with have varying coverage and thus I definitely have a wide ranging ability to produce high quality / complete mags. So if I am unable to build mags, could I still do gene predictions based on just the scaffolds? If so, in the atlas framework, would this be just the quality control, assembly, and annotation? Is this an appropriate way to estimate functional content?

Any advice you can provide would be much appreciated!

[SilasK]

Hello,

It is true the main usage of atlas is to generate MAGs of a dataset. You are afraid that you will not be able to produce high-quality bins from your dataset, whichz is understadable. Some metagenomes are difficult.

Just to keep in mind that atlas creates a non-redundant set of MAGs (one per species). So you only would need 1 good quality genome per species from any of your samples.

So for many metagenome this should be worth to try how many reads are mapped with the MAG aproach atlas run genomes

• There is a technical limitation that a sample without any bin is stuck. But we are working on solving this.

[SilasK]

However the good news is even if you don't get anough MAGs, there is the genecatalog workflow atlas run genecatalog It produces a non redundant set of all genes infered from a assembly. This should be what you are looking for.

For now, the genecatalog is annotated using eggNOG mapper. WggNOG mapper has some links to Kegg. This should be ok to calculate a functional richness/diversity.
But we are also planning to annotat the genecatlaog directly with the KOfams.