What is the most accurate approach for de Novo sequencing?

Question

I'm trying to decide between PacBio HiFi or Illumina sequencing platforms for sequencing the genome of aChrysina scarab. We want to identify the color pattern loci and also perform a phylogenetic analysis as we'll be sequencing several species of that genus.

When considering a hybrid approach I was told by someone at a sequencing facility that adding short reads to polish the HiFi reads assembly lowers the quality of the assembly and introduces more errors that corrections. Why is this? I haven't been able to find this statement in the literature.

Answer 1

HiFi reads can resolve most repeats in a genome. Illumina reads from these repetitive regions are likely to get mismapped and these mapping errors may become consensus errors. Most short-read polishing tools developed in the old days (e.g. pilon) will reduce the contig consensus quality. Furthermore, when you have a non-inbred diploid genome, HiFi reads can locally phase a large portion of the genome, while Illumina reads are likely to cancel/discard most HiFi phasing information, which would lead to a worse and less informative assembly.

In theory, if you use Illumina reads the right way, you can further improve HiFi contig consensus like what the telomere-to-telomere team is doing here. However, this pipeline is for haploid/inbred genomes only. There are no tools for generic genomes. Developing your own is beyond the capability of most people in the field. To this end, the person at your sequencing facility is mostly right: don't polish HiFi assemblies with Illumina reads.

Answer 2

Very interesting. Personally, I think the most important thing than anything I can input is it is worth asking about the HiFi quality index being considered and whether this was compatible with your biological objective.

Just to highlight that insect genomes are very different to genomics here - traditionally and even now - there has been a disproportionate focus on dipteran genomes. Okay certain Lepidoptera, namely the silk moth (Bombyx), is an advanced area of study. Beetles genomes are even more different still, agreed they holometabolous insects but still along way from dipterans. In addition, its worth pointing out beetles were late on the genome sequencing scene.

I don't know the precise biological objective - but I'm going to guess it. I'm assuming your wanting SNP variation to associate with phenotype colour variation, for the target genes (just guessing).

HiFi reads

With regarding mixing the genome sequencing approaches: what is the quality scale being used? If its lowering the chromosomal assembly (traditionally a key difficulty in insects), or weakening contig assembly then ok I could imagine. This is very different DNA. If the quality index is relating to lowering SNP accuracy I would be surprised, for others to comment, but just can't see it.

I am aware that a key issue is that insect genome, specifically with reference to dipteran genomes, has caused extreme difficult assembling due to the high AT content and repeat regions, with chromosomal identification a traditiona nightmare and requiring wet lab chromosomal mapping (Diptera). I think beetles are an outlier even on this spectrum - against other insects. The advice you might have been given could be with this issue in mind, particularly if it was from another entomologist.

I don't know if this genome has been sequenced, the implication of the question (title) is that is hasn't been sequenced yet and which leads to the second issue,

JBS Haldane

God has an inordinate fondness for stars and beetles.

There are an awful lot of beetles and a lot of those genomes not sequenced yet.

Personally I think you don't want accurate (really to maximise) genome assembly. In that scenario its got to be Illumina short-read, because this is a SNP study, where high accuracy calling from read depth is needed, given my guessed objective.

If the genome hasn't been sequenced there could be pressure to enhance genome assembly rather than target your biological objective. This could be why there has been mixed advice, between what's good for the community and what's needed for your biological objective, this being colour variation genes - for very pretty beetles.

I think, personally, its not necessarily a technical issue its either a win for others, or win for you. The only caveat is if the genes of interest are some extensive, physically linked gene family, then of course short-read may not be cool. I am not aware of that is a key feature of insect genomes.

---

In insect genomes there has been misplaced excitement ... new insect genome ... hang on thats just unordered contigs. This will happen using a short-read only approach.