Mistakenly purge of homozygous seqence in polyploid genome

[Axolotl233]

Hello Dr. Cheng and community,

I recently used hifiasm to assemble a allotetraploidy genome, and my command is hifiasm -o out -t60 --primary --h1 hic_1.fq.gz --h2 hic_2.fq.gz ccs.fastq.gz. HiFiasm was worked well and generated all output normally. The summary of hap1 assembly, hap2 assembly and primary assembly was here: (Total and N50):

hap1 assembly : 688M, 25M
hap2 assembly : 585M, 17M
primary assembly : 670M, 27M

Then I performed colinearity analysis between these three assemblies and potential diploidy progenitor genome. The result showed
obvious dual colinearity between primary assembly and diploidy progenitor genome in the most of region expect one chromosome. this region only has one homozgous sequece with progenitor genome in the primary assembly and other regions have two (marker with blue rectangle in primary.png).

I aslo checked hap1 and hap2, the same situation was observed in hap2 assembly but not observed in hap1 (marker with green and red rectangle respecitively)

I wonder if this region was truely existed in the one subgenome of our allotetralpoid, so I checked it with re-mapping methods. the result comfirm that this region is truely existed, and it was partially assembled in hap1, but not assembled well in primay and hap2. I guess this region has higher similarity (which need to further comfirm) than other genome region between two subgenomes because unknown reasons, which caused mistaken purge of hifiasm. Although I could use hap1 assembly directly in downstream analysis, but I still want to further improve assembly quality. I have tried to adjust parameter like add -s 0.75 or -l0, the former had no effect, and the latter produced a large number of chimeric sequences (between the two subgenomes) in the primary assembly. could you have any suggestion about it？

[Axolotl233]

I have tried to increase vaule of parameter D, a and r, the result are improved but still not completed in this region, could you have any progress about this question?

[chhylp123]

It actually is case by case. Would it be possible that you can share the bin files with me?

[Axolotl233]

Thank you, I have mailed my data file to you (haoyu.cheng@yale.edu), please check it