[hifiasm_0.25.0-r726] Single k-mer peak (16X) contradicts high heterozygosity ratio (289:1) in assembly

[darkxin12]

I am seeking guidance to resolve an apparent inconsistency in my analysis. While learning to use hifiasm for a genome assembly project, I noticed two results that I might be misunderstanding:

k-mer profile: The k-mer coverage distribution shows a single peak at ~16X ([M::ha_ft_gen] peak_hom: 16), which initially suggested a homozygous genome based on standard k-mer theory.
Heterozygosity statistics: However, the log reports a high heterozygosity ratio ([M::stat] heterozygous:homozygous bases ≈289:1), which typically implies a highly heterozygous genome with bimodal k-mer peaks.

[M::ha_analyze_count] lowest: count[5] = 11210960
[M::ha_analyze_count] highest: count[16] = 73404125
[M::ha_hist_line] 2: ********** 7517307
[M::ha_hist_line] 3: ******** 5950814
[M::ha_hist_line] 4: *********** 7923837
[M::ha_hist_line] 5: *************** 11210960
[M::ha_hist_line] 6: ********************* 15651122
[M::ha_hist_line] 7: ***************************** 21069458
[M::ha_hist_line] 8: ************************************** 27882266
[M::ha_hist_line] 9: ************************************************ 35166958
[M::ha_hist_line] 10: *********************************************************** 43039027
[M::ha_hist_line] 11: ********************************************************************* 50968341
[M::ha_hist_line] 12: ******************************************************************************** 58504231
[M::ha_hist_line] 13: ***************************************************************************************** 65161626
[M::ha_hist_line] 14: ************************************************************************************************ 70143067
[M::ha_hist_line] 15: *************************************************************************************************** 72966931
[M::ha_hist_line] 16: **************************************************************************************************** 73404125
[M::ha_hist_line] 17: ************************************************************************************************* 71437531
[M::ha_hist_line] 18: ******************************************************************************************** 67359635
[M::ha_hist_line] 19: ************************************************************************************ 61349348
[M::ha_hist_line] 20: ************************************************************************** 54563791
[M::ha_hist_line] 21: ***************************************************************** 47861182
[M::ha_hist_line] 22: ********************************************************* 41575691
[M::ha_hist_line] 23: ************************************************* 36265920
[M::ha_hist_line] 24: ******************************************** 32037788
[M::ha_hist_line] 25: *************************************** 28721116
[M::ha_hist_line] 26: ************************************ 26486023
[M::ha_hist_line] 27: ********************************** 25044742
[M::ha_hist_line] 28: ********************************* 24079502
[M::ha_hist_line] 29: ******************************** 23632477
[M::ha_hist_line] 30: ******************************** 23467603
[M::ha_hist_line] 31: ******************************** 23153048
[M::ha_hist_line] 32: ******************************* 23015716
[M::ha_hist_line] 33: ****************************** 22380247
[M::ha_hist_line] 34: ***************************** 21389297
[M::ha_hist_line] 35: **************************** 20376062
[M::ha_hist_line] 36: ************************** 19146936
[M::ha_hist_line] 37: ************************ 17626954
[M::ha_hist_line] 38: ********************** 15984342
[M::ha_hist_line] 39: ******************** 14378007
[M::ha_hist_line] 40: ***************** 12787803
[M::ha_hist_line] 41: *************** 11132014
[M::ha_hist_line] 42: ************* 9605896
[M::ha_hist_line] 43: *********** 8143294
[M::ha_hist_line] 44: ********* 6926763
[M::ha_hist_line] 45: ******** 5801225
[M::ha_hist_line] 46: ******* 4850012
[M::ha_hist_line] 47: ****** 4048750
[M::ha_hist_line] 48: ***** 3410010
[M::ha_hist_line] 49: **** 2845662
[M::ha_hist_line] 50: *** 2418978
[M::ha_hist_line] 51: *** 2045373
[M::ha_hist_line] 52: ** 1798807
[M::ha_hist_line] 53: ** 1573588
[M::ha_hist_line] 54: ** 1425472
[M::ha_hist_line] 55: ** 1288114
[M::ha_hist_line] 56: ** 1208866
[M::ha_hist_line] 57: ** 1117988
[M::ha_hist_line] 58: * 1034859
[M::ha_hist_line] 59: * 973983
[M::ha_hist_line] 60: * 928617
[M::ha_hist_line] 61: * 856342
[M::ha_hist_line] 62: * 814749
[M::ha_hist_line] 63: * 773728
[M::ha_hist_line] 64: * 733380
[M::ha_hist_line] 65: * 684491
[M::ha_hist_line] 66: * 661488
[M::ha_hist_line] 67: * 630514
[M::ha_hist_line] 68: * 609039
[M::ha_hist_line] 69: * 575526
[M::ha_hist_line] 70: * 555132
[M::ha_hist_line] 71: * 535845
[M::ha_hist_line] 72: * 517110
[M::ha_hist_line] 73: * 497096
[M::ha_hist_line] 74: * 473363
[M::ha_hist_line] 75: * 449207
[M::ha_hist_line] 76: * 429007
[M::ha_hist_line] 77: * 410004
[M::ha_hist_line] 78: * 400178
[M::ha_hist_line] 79: * 385339
[M::ha_hist_line] 80: * 375382
[M::ha_hist_line] rest: ************************** 19199508
[M::ha_analyze_count] left: none
[M::ha_analyze_count] right: none
[M::ha_ft_gen] peak_hom: 16; peak_het: -1
[M::ha_ct_shrink::930.6807.59] ==> counted 19574890 distinct minimizer k-mers
[M::ha_ft_gen::943.417[email protected]] ==> filtered out 19574890 k-mers occurring 80 or more times
[M::ha_opt_update_cov] updated max_n_chain to 100
[M::yak_count] collected 1440731874 minimizers
[M::ha_pt_gen::1164.312*11.07] ==> counted 83289253 distinct minimizer k-mers
[M::ha_pt_gen] count[4095] = 0 (for sanity check)
……
[M::stat] # heterozygous bases: 3820503823; # homozygous bases: 13212505

Could you help me understand:

How these observations might coexist under hifiasm’s internal model?
Whether adjustments to parameters (e.g., --hom-cov) could reconcile this discrepancy?

I would greatly appreciate any insights to improve my understanding and ensure proper assembly configuration.

hifiasm2.log

[olekto]

Hi,
you have a nice shoulder around 30-32x coverage, so I would not classify this distribution as "single peak". It is rather quite heterozygous (with the homozygous k-mers at 32x or so) as you say.

[darkxin12]

Hi,
Thank you for your insights. I noticed that Hifiasm printed:

[M::purge_dups] homozygous read coverage threshold: 29

Since this is close to the homozygous coverage we discussed, do I need to reset the --hom-cov value and rerun Hifiasm, or is the current setting sufficient?

Best regards!

[olekto]

Hi,
I don't think you need to do anything unless you get an assembly that is different to what you would expect (whatever that could be).

Ole

[darkxin12]

Thanks for the suggestion!
In our current phased assembly (haplotype-resolved assembly), the primary contigs (p_ctg) exhibit a 20% size increase compared to the last version of reference genome (Ref: 1.6 Gb vs. Assembly: 2.0 Gb).
Specifically, haplotype 1 (hap1) approximates the reference size at 1.6 Gb, whereas haplotype 2 (hap2) reaches 2.0 Gb.
I am testing smaller values for -s.