Evaluation of PacBio only de novo assembly

Question

[lkw222]

Does anyone have experience using validation/evaluation methods (other than basic statistics like N50) for a PacBio only de novo assembly? What would you recommend? Note: I don't have multiple assemblies to compare, just want to try to do some validation of a single de novo assembly.

Answer 1

[lexnederbragt]

There is not a dedicated tool to do this. What comes to mind is mapping the reads back to the assembly and calling variants, especially structural variants. Tools like PBHoney, Sniffles and the like. The problem is that only looking at one assembly won't tell you much - some structural variants may diploid (if your organism is diploid) rather than assembly errors.

Finally, if you expect a lot of synteny to another genome for which there is a high-quality reference, you can compare your assembly to that reference, e.g. using GenomeRibbon.

Answer 2

[Rohit]

It would be best to start to see how much of the assembly is actually meaningful through single-copy orthologs comparison. Check out CEGMA or BUSCO.