MEGAN and Kraken for taxonomy classification

Question

[deepthithomaskannan]

Hi,

I am trying to do taxonomy classification for my unmapped (to reference genome) reads. I tried both Megan and kraken.

In kraken I can use read files directly as the input, whereas in Megan I have to make the blastN file first from the read files to give it as the input.

No plotting tool is readily available for kraken, while Megan already has a plotting tool.

I used bacteria,virus and fungi genomes as reference database. I am trying to figure out if I can use read files directly as an input to MEGAN so that I can skip the BLASTN step.

I also want to compare the Megan and kraken output.

Thanks,
Deepthi

[Ram]

Which BLASTN step? Please provide some context.

Comparing your two other questions, I think you would be better off if you'd made a single question explaining your premise, what you are looking to achieve and then listing your questions in a specific fashion with relevant context.

EDIT: Restructured your question so it is easier to compartmentalize while reading.

[ramanandan.prabhakaran]

Hi Deepthi,

I am also working on kraken to build my own custom DB. As you mentioned in the above post, you built your database consisting of bacteria, virus and fungi. Even my requirement is to build custom DB with the same three species.

1. Can you help me with that custom DB?
2. Can I use RefSeq proteins instead of RefSeq genomes to build custom DB?

Cheers,
Ram

[Phil S.]

Hi,

kraken can only use AGTC sequences, everything else it will treat as ambigous. So no, you can't you proteins (afaik). The creation of custom databases is really nice explained in the kraken manual (which is, in its whole, a nicely written thing!). The manual can be found here

HTH,

phil

[ramanandan.prabhakaran]

Thanks Phil for providing me a nice explanation.

I am new to metagenomics field. I would like to generate synthetic reads similar to 454 or Illumina. I figured out MetaSim can generate it. But I have doubt, can we use gene sequences (fasta format) from NCBI as a input to generate synthetic reads instead of genome sequences?

Answer 1

[Phil S.]

Hi,

it is not possible to directly input read files into megan, afaik. However, to get a first guess what your reads are depicting using kraken you can use krona-tools (which they also used) in their paper to visualize results.

HTH

[deepthithomaskannan]

Thank you Phil.

Deepthi