This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Metagenomic classification by custom database

Hi,

I'm looking for a metagenomic classifier which will classify paired end illumina reads using a custom fasta-file as a database. I have been looking at a bunch of them and they all are very concerned with taxonomy, which is not relevant - or exists - in this particular case (Centrifuge is doing a fine job of that).

Ideally the output is a table of abundance for each entry in the database.

Any such thing?

sequencing classification
• 3,685 views • link
updated by Carambakaracho Carambakaracho • written by mikael.lenz.strube • mikael.lenz.strube

2 answers

To me it sounds all you need is some super fast blast like functionality. In case I didn't misinterpret your request try either Benjamin Buchfink's diamond (against protein database) or NCBI magicblast (against DNA database).

The abundance table then is a simple script using hash/dictionaries or something similar.

• 1 views • link
written by Carambakaracho Carambakaracho

Hi Mikael, fyi, general good practice is to mark helpful answers with thumbs up and in case it was the correct answer with accepted answer, so others can spot the best solutions faster

• 1 views • link
written by Carambakaracho Carambakaracho

You could try Kaiju. Check the github page here

Read under custom database section

• 1 views • link
written by lakhujanivijay lakhujanivijay

Hi Vijay,

I looked at Kaiju, but it requires NCBI taxon identifiers for custom databases, which i don't have.

• 1 views • link
written by mikael.lenz.strube • mikael.lenz.strube

Hi genomax,

the issue is that i have no taxonomy in my own database, they potentially have arbitrary and anonymous headers.

• 1 views • link
written by mikael.lenz.strube • mikael.lenz.strube

Hey, I just checked it again and it says that it doesn't need the taxonomic classification. So may be you can give it a try.

• 1 views • link
written by harish harish

I'm not sure, it says

It is also possible to make a custom database from a collection of protein sequences. The format needs to be a FASTA file in which the headers are the numeric NCBI taxon identifiers of the protein sequences, which can optionally be prefixed by another identifier (e.g. a counter) followed by an underscore, for example:

Am I misunderstanding something?

• 1 views • link
written by mikael.lenz.strube • mikael.lenz.strube

Log in to answer this question.

More posts like this