Assemblers for short read bacterial isolates data

Question

[shevch2009]

Hello all, I am looking for assemblers for shotgun short reads data from bacterial isolates (permafrost). There are many articles where assemblers use Nanopore and ILLUMINA data, but I only got ILLUMINA data. I tried spades and megahit, but using kraken taxomony showed that one isolate have a few dozen different species, although 68% no hints. This is really weird, sure it could be a contamination, but is there anything else I could try? Should I maybe do a binning? This is my first time working with this type of data. I would appreciate any ideas! Thanks. Alla

[andres.firrincieli]

but using kraken taxomony showed that one isolate have a few dozen different species, although 68% no hints

Different species of the same genus? If that's so, I would not worry too much. Go ahead with the assembly and calculate the contamination levels with checkM2 or similar

[shevch2009]

No, different families, genera, that is what makes it strange.

Thanks)

Answer 1

[Mensur Dlakic]

Both SPAdes and MEGAHIT do a good job with Illumina data. One potential problem could be if you have a coverage that is too deep say > 1000x. If you have deep coverage, digitally reducing it to 200-300x might produce an assembly that behaves better.

If there are different species in your assembly, binning should easily separate them. However, it won't work nearly as well for the strains that share very high identity.

[shevch2009]

Thanks) We had 100 coverage, should be ok for bacterial isolates. We have different families, genera, instead of one. Not really sure what can I do.

[Mensur Dlakic]

Binning can be done in seconds, and there is no harm in trying it. If you run CheckM analyses before and after binning, it should give you a clear answer whether your microbes are different enough in terms of tetranucleotide frequencies to separate into distinct groups.