search

chhylp123 / hifiasm

Hifiasm: a haplotype-resolved assembler for accurate Hifi reads
MIT License
526 stars 86 forks source link

Possible missing one haplotype in human assemblies #635

Open woodoo46 opened 5 months ago

woodoo46 commented 5 months ago

Hi there,

We are using hifi-asm to assemble human genome sequenced on Pacbio Revio machine, the coverage of the sample is roughly 30X (may be lower though), we found that hifi-asm might miss one haplotype in some low coverage region, eg the sample have two haplotypes, but the assembler reports the region only in one set of contigs. Is there any parameter setting (which I may miss), or internal hard-coded setting I can change, so the assembler can output contigs as aggressively as possible in both sets of haplotypes?

Thanks.

George

chhylp123 commented 5 months ago

Would you mind showing some general numbers or specific examples for that? Then I could figure out what happened. Thanks so much!

woodoo46 commented 5 months ago

I am not sure what I should give to you. One thing is one haplotype may be sequenced a little more than the other according to pacbio haplotagged bam file. We think there are two haplotypes is because there are noticeable number of heterozygous calls from the region (deepvariant calls).

Or if there is any log information we can check that anything has been dropped?

Thanks.

George