Bacterial ITS analysis

[SDPA_Pet]

Hello, I was wondering if anyone has done Bacterial ITS (not fungal ITS) community diversity analyses before? I do 16S rRNA all the time using Qiime or Mothur. However, the two software doesn't support bacterial ITS. If anyone has done it before, can you tell me your analysis work flow?

Thanks
Ben

[rhinoceros]

However, the two software doesn't support bacterial ITS.

Sure they do, however, afaik there's no reference for bacterial ITS so you'd be doing just de novo clustering. No idea how you'd go about assigning taxonomy to your reads. Blastn against a very sparse collection perhaps?

[SDPA_Pet]

HI rhinoceros,

Thanks. "Blastn against a very sparse collection". Do you mean BLASTn against a small and specific database. I know ncbi has mircrobial rRNA database. I am not sure if this database only has 16S or includes ITS regions. It is obvious that ncbi NT database is too big for me.

Any suggestions?

Ben

[rhinoceros]

HI rhinoceros,

Thanks. "Blastn against a very sparse collection". Do you mean BLASTn against a small and specific database. I know ncbi has mircrobial rRNA database. I am not sure if this database only has 16S or includes ITS regions. It is obvious that ncbi NT database is too big for me.

Any suggestions?

Ben

NCBI hosts a 16S db, which doesn't include ITS seqs. Bacterial ITS sequencing has never been a thing, so a bacterial ITS reference db would be limited to the few fully sequenced bacterial genomes, hence "sparse" and pretty much useless..

[SDPA_Pet]

NCBI hosts a 16S db, which doesn't include ITS seqs. Bacterial ITS sequencing has never been a thing, so a bacterial ITS reference db would be limited to the few fully sequenced bacterial genomes, hence "sparse" and pretty much useless..

Do you mean if I do BLASTn search, I should use NCBI nr database instead of nt database?

I know NCBI BLASTn web portal can search nr and nt database at the same time? Can BLAST command line BLAST against two databases at the same time?

[rhinoceros]

Do you mean if I do BLASTn search, I should use NCBI nr database instead of nt database?

I know NCBI BLASTn web portal can search nr and nt database at the same time? Can BLAST command line BLAST against two databases at the same time?

What I mean is that you should forget about the whole thing. Anyway, nr is a protein db. The default NCBI's web blastn is against non-redundant version of nt. But many sequences in nt (or nr for that matter) don't have any species annotation other than "uncultured bacterium/archaeon/organism". What good would that information be to you? Optimally, you'd build an ITS subset of e.g. refseq_genomic, but again, this db wouldn't have that many sequences, so it would be a rather miserable reference db.

[SDPA_Pet]

What I mean is that you should forget about the whole thing. Anyway, nr is a protein db. The default NCBI's web blastn is against non-redundant version of nt. But many sequences in nt (or nr for that matter) don't have any species annotation other than "uncultured bacterium/archaeon/organism". What good would that information be to you? Optimally, you'd build an ITS subset of e.g. refseq_genomic, but again, this db wouldn't have that many sequences, so it would be a rather miserable reference db.

OK. Thanks. Just double check with you this,

The default NCBI BLASTn database on the web is nucleotide collection (nt/nr).

So the nt means nt database
and nr means non-redundant version of nt?

Yes, I understand it is very hard to get exact species name for microbes even using 16S rRNA gene marker.

[rhinoceros]

OK. Thanks. Just double check with you this,

The default NCBI BLASTn database on the web is nucleotide collection (nt/nr).

So the nt means nt database
and nr means non-redundant version of nt?

Yeah, it's a special non-redundant version of the nt db which they distribute via their ftp. In their ftp, the "main" protein db is called just "nr".

Yes, I understand it is very hard to get exact species name for microbes even using 16S rRNA gene marker.

Well, yeah, that's usually the case. However, as I'm sure you're aware, there's no consensus of what makes a prokaryote species, and anyway, genus level annotation is usually good enough. If you want more specific than that, you'd be looking into phylogenomics instead..