How Do You Analyze Metagenomics Data For Species Content And Biological Function?

Question

[Larry_Parnell]

We will have fecal samples from humans fed different diets (mainly differing in carbohydrate content and type) and those samples will be subjected to whole-genome, shotgun sequencing. This is more in-depth and more broadly focused than ribosomal DNA sequencing to determine species (or genus) present. I have read related BioStar questions regarding arriving at estimates of diversity, enrichment methods and the basic process flow using MEGAN. But I have a different question.

Because we would like to know which species are present as well as specific gene content in order to determine metabolic potential of the gut microbiome (so that we can assess what happens to the ingested carbohydrate), the choice of analysis tool(s) becomes important. Thus, which analysis tool for describing gene content do you prefer, MEGAN, MG-RAST or another? Do you prefer to use as input the sequencing reads or their corresponding BLAST output?

Any advice from the community on this topic will be appreciated.

Answer 1

[Michael]

While Megan and MG-RAST seem to be the most commonly used tools, I would also have a look at the CARMA-pipeline, which uses Environmental Gene Tags (EGTs) in conjunction with a PFAM search, which I believe is a bit more advanced than the relatively straight-forward blast approach of Megan and RAST (using the SEED database). Also, the CARMA3 paper suggests that it outperforms Megan and SOrt-ITEMS. There is also the WebCarma version, on that page you also find the source code.

The idea about using EGTs and the algorithm of CARMA is, imho, much better described in the original paper of the inventors than in the later derivative versions.

Hope this helps, disclaimer: atm I am not actively working with meta-genome data myself.

CARMA3 paper

[Larry_Parnell]

Thank you, Michael. This is exactly the type of insight I was hoping to receive. We're new to this and all info helps.

[Wolfgang]

CARMA1 and CARMA2 use the same method which is described in the original CARMA1 paper. CARMA3 uses a completely new method that has little to do with the original method. This new method outperforms MEGAN, SOrt-ITEMS and the older versions of CARMA. The WebCARMA paper used CARMA2, but in the meantime WebCARMA has been updated to CARMA3.

[Michael]

Hi Wolfgang, thanks for the clarification. I think I got confused with the versions of the algorithm here.