Tools for metagenomics classification using long reads

Question

[Medhat]

Searching for taxonomic sequence classifier tools that assigns taxonomic labels to reads, where the input is long reads (PacBio, ONT .....).

There is already tool that could work with long read (centrifuge). But I would like to know if there are any other tools beside centrifuge?

Thanks

[cschu181]

kraken? Although, with PacBio you might have a problem with erroneous k-mers.

[Medhat]

Kraken supports short read as far as I can tell, beside that Centrifuge is faster

[cschu181]

I don't think long reads would be an issue, it would just treat it as single-end fastq. When I look at Table2 in the centrifuge paper, kraken is about twice as fast as centrifuge, albeit with a much worse memory requirement (and of course the database creation takes ages).

[Medhat]

Do not forget that long read comes with high error rate (15% for PacBio as example).

[cschu181]

That's what I wrote in my first comment. I don't know about ONT, but of course Pacbio will introduce erroneous kmers due to their indel error model.

Answer 1

[WouterDeCoster]

From Oxford Nanopore there is WIMP. I have never used it though. See also this link. I guess you need a nanopore community account, and can then access the tools here. I actually don't even know if it's freely available.

[Medhat]

It is not free, and the result is more general than Centrifuge.

Answer 2

[predeus]

It seems like MetaMaps should be quite good for this task. I'm about to test it with a couple of my datasets.

[Medhat]

Please keep us updated wih your benchmark, and if it give some ouput that could be represented in Krona tools?

Answer 3

[bioinfo17]

MetaMaps is pretty good for long reads - can go up to strain-level taxonomic assignment.