Filter human parasites from taxonomic profiling with Kraken

Question

[Fungsten]

I am trying to generate a summary of parasites in human metagenomic samples.

I have been looking into Kraken2 and its databases built from Refseq (which should contain all parasite sequences annotated). Have also noticed this DB available from EuPathDB:

https://ccb.jhu.edu/data/eupathDB/

Is it a good approach to solely use EupathDB or should I be looking into other DBs for any missing parasites that might not be included there?

Answer 1

[Asaf]

I personally think that if you're looking into parasites specifically your better option is to have a protein dataset and diamond blastx your DNA against it. I think that it will yield better results than using k-mer based methods.

[BioinformaticsLad]

Could you explain why? Do parasites tend to have higher mutation rates?

[Asaf]

Me today is trying to understand what me six months ago meant. I think that it's true for everything, not just parasites.

Answer 2

[ctseto]

Using CAT (which in turn uses diamond), and then parsing its contig2classification output files by taxonomy may be helpful; everything with a taxonomy tree that descends towards parasites could then be eliminated.