mixed-ploidy data (haplodiploid insects): Can Stack2's bayesian algorithm handle it, or better off using Stacks1?

[Claude Patrick Millet]

Hello all,

This is a re-post (after deleting) of a previous question I posted back in November 2025, which went unanswered. I have a more specific question related this topic, which is why I am reposting it under a less vague title.

I am gearing up to analysis RADseq data (paired-end, EcoRI-cut) from a parasitoid wasp species. I know that among my data are males, females, and samples of unidentified sex. In many hymenopterans including my species, males are haploid and females diploid. I am of course intending to perform post-VC analyses separately for males and females. My plan is however to perform de novo SNP calling using Stacks on all my samples, then use differences in heterozygosity (which should be zero in males, save for errors) to sex my unidentified samples. 

Also, given the mixed-ploidy of my data, I am wondering if there 

I was wondering if there were any counter-arguments or red flags around using stacks for variant calling on a mixed-ploidy sample set? I saw this previous discussion but

there isn't a whole lot in that thread. Anybody have experience trying such an analysis?

Also, given the mixed-ploidy of my data, I am wondering if there are any advantages for or against using Stacks1 or 2. As I understand, Stacks 2 will consider previously called genotypes when calling new ones, while Stacks1 will not. I was thinking given the mixed ploidy (and therefore two expected patterns of zygosity in my samples), it might be best to use Stacks1 which will treat each samples independently, lest Stacks2 's variant calling get "confused" by this.  But I'm not sure if this assumption is correct? Any thoughts?

Thanks!

Patrick