Getting highly imbalanced haplotypes using trio-binning

[amara86]

Hi
I am attempting to generate trio-binned assemblies using Hifiasm and have followed this workflow:

Processed Illumina HiSeq short reads from Parent 1 using fastp for adapter removal and quality filtering (Total: 304,817,131 bp).
Processed 10X Genomics linked-read data from Parent 2 using fastp (Total: 88,641,287 bp).
Collected HiFi reads from an F1 hybrid offspring for assembly.
To perform trio-binning, I generated parental k-mers using yak, but the number of unique k-mers was highly imbalanced:

Parent 1: 34,008,379 k-mers
Parent 2: 7,669,877 k-mers
This imbalance resulted in highly uneven haplotype assemblies:

One haplotype: ~948MB
Other haplotype: ~5MB
To address this, I tried generating k-mers using KMC, which yielded more balanced unique k-mers:

Parent 1: 308,155,996 k-mers
Parent 2: 280,131,731 k-mers
However, despite this, the assembled haplotypes remain highly imbalanced.

Questions:

1. Is there a way to normalize k-mer counts in yak to achieve a more balanced distribution?
2. Can we convert k-mer generated by (KMC in txt format) to yak format?
   Regards,
   Amara

[chhylp123]

Hifiasm accepts -3 and -4 if you can partition reads by yourself. If you can do it with KMC output, that might work.