Comparison samples prok-mainly vs mixed prok/euk

[HannahBenisty]

Hi SqueezeMeta team, I am running SqueezeMeta two closely related environmental metagenomes (same site), using --euk with the same workflow/settings (sequential mode). One sample is expected to be mostly bacterial, while the other is expected to be a mixed community with eukaryotic DNA plus bacteria. I see a large drop in contig-level taxonomy assignment in the mixed sample:

• Run A: 5542 contigs; 75.0% assigned; 25.0% Unclassified
• Run B: 5795 contigs; 29.1% assigned; 70.9% Unclassified

• Is this magnitude of drop expected when the sample contains substantially more eukaryotic DNA, even with --euk?
• Any recommended flags/steps to improve (e.g., -D doublepass)?

I have several paired samples like these ones, so it would be great to know if there would be a way to improve.

Thanks!
Hannah

[fpusan]

So if I understand correctly:

Run A corresponds to the mostly bacterial sample
Run B corresponds to the sample with eukaryotic DNA

You used --euk for both.

Is this correct?

[HannahBenisty]

Yes, that is correct. For run A we also expect eukaryotic DNA but less than in run B.

[fpusan]

Then -euk is correct (but you don't really need it if planning to use SQMtools later, since by default it will also ignore identity thresholds for the taxonomic assignment of eukaryotic contigs, see here).

-D will definitely help here, but maybe not by a lot.

I would say the drop itself is not surprising, the magnitude will depend on how many eukaryotes you have in there, and how well represented they are in the databases. Sometimes sqm_reads.pl will give better results for taxonomic classification of this kind of samples.

[HannahBenisty]

Thanks a lot! I will try -D.