Any suggestion on metagenomics pipelines for processing shotgun metagenomics whole genome sequencing samples?

Question

[eli_bayat]

Hi, I have some shotgun metagenomics gonorrhea samples and I want to use metagenomics pipeline to process the samples and then detect antimicrobial resistance. I looked into few pipelines so far for metagenomics data such as ATLAS, Sunbeam, MetaSUB/CAP2 and nf-core/mag. They have different steps of data processing with different features and tools used.

1. Is there a review paper or some documents that compares these pipelines in terms of pros/cons?
2. Is there any other pipeline(s) that based on your experience, I can use? or even any pipeline that you think works better among the ones I mentioned?

Any suggestions will be appreciated.

Thanks

[andres.firrincieli]

The ultimate choice on which pipeline is better depends on your needs. For example, since I am mostly interested in MAGs I use metaWRAP because it has a module called Bin_refinement which produce a final bin set from the output of three different binning algorithms. Of the pipelines you mentioned I would avoid the following:

• nf-core/mag: it use only one binning algorithm (metaBAT2).
• MetaSUB/CAP2: the assembly module consist only in one assembler, metaSPADES which is computationally more expensive when compared megahit
• Sunbeam: I would not pick this one only because it does not have a binning module. However, if you do not care about MAGs and are interested only in antimicrobial resistance genes then Sunbeam would work just fine

[eli_bayat]

The points you mentioned are important to consider and I haven't thought about them. Thank you so much for pointing these out. very helpful!

[pfchuckran]

Not aware of any paper comparing different workflows, but I will add one more to your list and suggest Anvi'o (publication: https://www.nature.com/articles/s41564-020-00834-3). I've been using it to assemble MAGs and have found the tutorials and documentation very approachable. They also have good support and an active community.

[eli_bayat]

Thanks for the suggestion. I just looked at the paper and it seems a lot of works were put on it. I like that it is also powered by Snakemake. Thanks!