Haploidify in ALLPATHS-LG

[will.jackson]

Hi,
Could someone please explain to me what the HAPLOIDIFY function does in ALLPATHS-LG? I have done two assemblies with ALLPATHS-LG - one with HAPLOIDIFY on and one with it off. The resulting assemblies are quite different - for example, when I use HAPLOIDIFY the N50 of contigs goes to 17kb as opposed to 5.4kb with it off.

Thanks very much

Will

[lorendarith]

The HAPLOIDIFY option is typically used when dealing with diploid data sets. It "hapoidifies" the differences in the diploid data set which makes it easier to assemble, then puts all the snp info in the resulting consensus back in the resulting efasta/fastg.

Haploidify is 'experimental' and is only really helpful if your genome is particularly polymorphic. Typically we found it useful if the computed polymorphism rate was 1 in 400 or higher. You are unlikely to find significant improvements for lower polymorphism rates and it could potentially even harm the assembly.

Taken from the user forum.

[will.jackson]

Hi lorendarith,
Thanks for the quick reply. I am working on a diploid genome and have a polymorphism rate of 1/118, so HAPLOIDIFY looks like a good option. Could you explain a little further what the option actually does please? What do you mean when you say it "'haploidifies' the differences"?
Thanks

Will

[lorendarith]

Sorry, but I can't help you more than that, little is known what the option really does or how it actually works.

If you have a quite polymorphic genome and use the HAPLOIDIFY option, it will probably collapse certain polymorphic regions instead of outputting them as separate contigs/scaffolds. It should therefore decrease the number of sequences and an increase in N50 can be observed. At least that is what others think and have seen.

I'm also working on a diploid genome, but my SNP rate is 1/862 and I can't say I've seen any major improvements when I used the HAPLOIDIFY option, just by looking at the metrics.

Try running an assembly with, without it and then compare. You should also then assemble less than the predicted genome size (K=25), when the option is employed.