# How to improve a genome assembly using Dovetail and PacBio assembly?

I have more of a conceptual question. I have two genome assemblies from the same plant, one from Dovetail technology (~998 Gb) and another is PacBio HiFi assembly (~1.1 Gb). The Dovetail assembly is more contiguous but has lower base quality whereas the PacBio has higher base quality but is more fragmented. It is a diploid organism with relatively high heterozygosity.

I tried to use RagTag patch to improve the assembly. But no improvement was obtained.

Is there a way to use both assemblies and produce a hybrid assembly with high contiguity and base quality?

• You can't do assembly with Dovetail alone. You have to have WGS reads in the first place. Is the Dovetail assembly derived from HiFi assembly? Jan 21 at 22:07

Answer from @maximilian-press, converted from comment:

I am confused when you talk about base quality, because those technologies are largely about ordering and orienting existing contigs, not assembling contigs in the first place. If you have Hi-C reads, you can read them into juicebox and actually visualize the junctions to see if the assembly makes sense: aidenlab.org/juicebox. It would also be helpful to see the actual numbers, not just "more" or "lower".

• But do not you think that it is possible to use the Dovetail assembly to scaffold the PacBio HiFi assembly with more contiguity? Jan 24 at 9:20
• @AnikDutta I think that the specific purpose of "Dovetail" (which I assume means Hi-C and/or Chicago) is the scaffolding of genomes. but on its own it is not much good for assembly. A "Dovetail" assembly is probably a set of contigs created by non-Dovetail technology (like PB HiFi), which have been scaffolded using an extra layer of "Dovetail" data. To be able to answer this question we would need to know what that non-Dovetail technology is that underlies the final assembly. Jan 25 at 17:16
• @AnikDutta For example: are the PB HiFi contigs the basis of the "Dovetail" assembly? If so, then the "Dovetail" assembly probably already represents an effort to scaffold the HiFi assembly. Jan 25 at 17:19
• @MaximilianPress Thanks a lot for the reply. No, the PB HiFi contigs are not the basis of the Dovetail assembly. As far as I know, Illumina reads were used for Dovetailing. The PB was generated independently. I ran the BUSCO and found out that the PB HiFi gives 98% BUSCO score whereas the Dovetail gave 95%. What do you think of using the Hi-C data and HiFi data to create haplotype resolved assembly? Will it make sense here considering it is a highly heterozygous diploid species? Jan 26 at 10:34
• @AnikDutta If the DT assembly is Illumina-based, I would strongly recommend the workflow of HiFi + Hi-C assembly. hifiasm natively takes in both of these data types: github.com/chhylp123/hifiasm, it should be pretty easy to run as long as you have the compute resources. You can polish the assembly post hoc using juicer/juicebox, after mapping the Hi-C reads onto the resulting assembly, as suggested above. Jan 26 at 18:27