How to determine contig taxonomy from metagenome?

Question

[A_heath]

Hi all,

I have a large metagenome that I know contains multiple species. Moreover, I have one contig in particular (within that metagenome) that I'm very interested in, and I would like to know what species it belongs.

What strategy would you use?

If you have suggestions, I'll gladly take them.

Thank you in advance!

Answer 1

[andres.firrincieli]

You can use Kaiju (https://github.com/bioinformatics-centre/kaiju/issues/73) or MEGAN-LR. If the contig is not too long you could try a simple Blast

[A_heath]

Thank you for your suggestions andres.firrincieli. I'm going to try it.

Answer 2

[zhangdengwei]

CAT, based on protein classification, is an alternative for your problem. But in my experience, it is hard to annotate relatively short contigs as well as a deeper resolution such as species level.

[A_heath]

Thank you zhangdengwei for your help. I will also tried CAT but I saw that the database was very large (>140 GB).

[zhangdengwei]

Yes, it indeed has an enormous database.

Answer 3

[lagartija]

Kaiju is a good solution for large databases but for your individual contig you should definitely do a blast against refseq or nr and download the taxids and see if they converge. If you want to do it locally you can try to blast it or use diamond against uniref90 (the database is smaller so you can actually have it locally)