How to calculate organism relative abundance from Illumina metagenomic data?

Question

[nkuyfq]

I have got pair-end metagenomic data (2 x 100bp) generated by Illumina Hiseq 2000 of several samples. Human sequences have been removed and I aligned the rest reads to bacteria reference genomes. I want to calculate organism content (namely, relative abundance) of each sample so that I can use this information in further comparative metagenomics analysis.

The question is how to calculate organism relative abundance from the rest reads. Note that different samples may have different sequencing levels, normalization may be needed. I need organism relative abundance that could correct for different sequencing levels in these samples. Is there a software which can handle this?

p.s.:Qin's paper entitled 'A human gut microbial gene catalogue established by metagenomic sequencing' talks about how relative abundance was calculted, however, there is no much detail provided. It's quite confusing.

As I am building my own analysis pipline, online resources might not fit for me.

[JC]

You can try MEGAN or MG-RAST.

[nkuyfq]

Thanks for your reply.

1. MEGAN seems to give absolute read counts to me, and I don't know whether it has corrected for different sequencing levels in my samples.
2. As I am building my own analysis pipline, online resources might not fit for me

[JC]

I see, maybe you can use QIIME or Kraken.

[nkuyfq]

Thanks, I will have a try.

[nkuyfq]

Both Kraken and Metaphlan seem good. I will try them out.

[nkuyfq]

Metaphlan seems only cover Bacteria and Archaea, which is a limiting factor for its use in Viral and Fungal metagenome.

Kraken is likely to be a resource-intensive software, but it's quite fast.

[deepthithomaskannan]

Hi,

I am also doing same kind of analysis. What do you mean by "different samples may have different sequencing levels"?

Thanks,
D

Answer 1

[Brett Kennedy]

Check out Kraken:

http://genomebiology.com/2014/15/3/R46

https://github.com/DerrickWood/kraken

Recently published out of the Salzberg lab. It's easy to run and install, if a bit resource intensive. I would be skeptical of anything mapping below species, but to the genus level it's reasonably accurate. Not to mention blazing fast.

Edit:

Kraken won't deliver immediate relative abundance, but an identification for each reference, which will likely recaputulate the relative abundance taken together.

MetaPhilAn is another tool that is specifically for relative abundance and is worth a look.

http://huttenhower.sph.harvard.edu/metaphlan

[Andreas]

+1 for metaphlan

Answer 2

[Len Trigg]

You can also try the metagenomics functions from Real Time Genomics. The RTG species tool takes SAM/BAM mapped to a reference species database and calculates abundances either by abundance of dna or by organism (taking genome lengths into account). If your reference includes taxonomic information (RTG provides pre-built ones that include bacteria and viruses) this is also taken into account, and reporting is both via tsv and krona visualization. (RTG Core is available free for non-commercial academic use).