Mg-rast api

[sfranzenburg]

Hi,

I have uploaded my metagenome files to MG-RAST. 30% of my reads are bacterial, which is what I expected.

How can I extract all bacterial reads only? I want to perform all my downstream analyses (e.g. PCA) only with the bacterial metagenome and discard all reads that hit the host species or viruses.

I guess I can use the API for that, however, I have no idea how to use it.

thanks,
Soren

[sfranzenburg]

No one?

How can you compare two different samples of host-associated metagenomes when they contain a huge amount of host reads? Of course they will be different, of course they cluster by host-species, but not necessarily because their microbiome is different.

So, what is the way to analyze host-associated metagenomes with focus only on the bacteria?

[roliwilhelm]

Same Question

Hi!

I wish I had an answer for you, but I am also looking for similar assistance. In my hunt, I've discovered PhymmBL, but it is only a different robust way for predicting taxonomy of short reads. This may be a more accurate method, but I still don't know what downstream measures one might take to isolate reads of specific taxonomy.

Did you figure out a method or find a tool?

Thanks,

[fibar]

Hello,
Good question, Soren.
My suggestion is to learn how to use the API, because one way to solve your problem is by downloading annotations directly. This is how I use it in the linux command line:

curl -X GET -H "auth: ... (key) " "http://api.metagenomics.anl.gov//matrix/function?id=mgm4603523.3&id=mgm4603525.3&group_level=level3&source=Subsystems&result_type=abundance&identity=60&filter_level=domain&filter=Bacteria" > subsystems_level3_R1-id60-2samples.bacteria.biom

Meaning: Retrieve abundance matrix of Subsystem annotations at level 3 for listed metagenomes at % identity > 60, filtered to return annotations only in domain Bacteria

From the API you can even retrieve data for ordination plots such as PCoA.

I hope this helps.

[sfranzenburg]

Hi fibar,
first, thank you very much for your answer, it helps.

It is just frustrating to see publications claiming two microbiotas are different based on PCA of MG-RAST data, which still include the host reads. Microbiota research took off so fast that alot of people do not know what they are doing and do in anyway... Bad for Science.