Which assembly should be used as a reference for population genomics: primary or haplotype-resolved assembly?

[hungweichen0327]

Hello,

Thanks for a great and useful assembler. We got wonderful results on different plant genomes.

I am assembling the diploid plant genome of using Hifiasm. According to the documentation, Hifiasm purges haplotig duplications by default and generates a primary assembly (p_ctg) together with two phased haplotype assemblies (hap1 and hap2).

I obtained the following assembly statistics:

The default primary assembly is slightly larger than hap.2 and has a slightly lower L90, whereas hap.2 contains substantially fewer contigs and has a similar N50 and completeness.

My goal is to select a reference genome for downstream population genomic analyses using WGS data from different individuals.

In this situation:

1. Would you recommend using the default primary assembly or hap.2 as the population genomics reference?
2. Is the primary assembly generally a better haploid representation because it is constructed specifically as the primary output after haplotig purging?

Thank you.