combine hifiasm and hifiasm-ont assemblies.

Question

[Mark]

I am wondering if it's possible to perform genome assemblies by combining PacBio HiFi and ONT data. How I understand it, contigs made by hifiasm from PacBio HiFi reads are more accurate but shorter, and contigs made by hifiasm-ont from ONT reads are more less accurate but more contiguous, I'm wondering if there's some tool that allows you to create de novo assemblies by creating a draft assembly from PacBio contigs, and then scaffolding them using ONT contigs. Does such a pipeline exist yet / is there a better, alternative strategy?

[GenoMax]

If you are working with a good model organism then you may be able to use available reference to try and merge the assemblies.

In case you have not seen this hifiasm author hints that a version that can use both PacBio and Nanopore data at the same time may be available in future: https://github.com/chhylp123/hifiasm/issues/849#issuecomment-3019737180

Answer 1

[shelkmike]

I suppose that you can just combine Nanopore and PacBio reads in one file and give this file to Hifiasm with the option "--ont". Probably, Hifiasm will treat PacBio reads just as a "more accurate fraction" of Nanopore reads.

Alternatively, you can make two separate assemblies and merge them with a tool like Quickmerge (https://github.com/mahulchak/quickmerge). However, in my experience, merging of assemblies in not a good thing, because if one assembly contains a structural misassembly, the merged assembly will have duplicated sequences. For example, if there is a genomic region A-B, and one of the assemblies contain these sequences in an improper order (B-A), then merging may produce A-B-A or B-A-B. In other words, some sequences will be duplicated. Such improper merges can be detected by an increase in BUSCO's "D".