Comparing homozygosity of k-mer plots

Question

Attached are two kmer plots from two closely related species. Is that safe to say that the one on the left has higher homozygosity than the one in the right k-mer plot, due to a low to almost flat blue peak (2x) in the left plot? My concern is that the blue peaks in the two plots are not in the same position: left plot at 4n, while right plot at 2n. I am aware of the grey area in the right plot indicating missing content in assembly, likely due to reads from multiple individuals (correct me if I am wrong).

Model for kmer plot on the left

Formula: y_transform ~ x^transform_exp * length * predict2_0(r1, k, d, 
    kmercov, bias, x)

Parameters:
         Estimate Std. Error t value Pr(>|t|)    
d       4.019e-02  1.262e-03   31.84   <2e-16 ***
r1      1.014e-02  1.664e-04   60.94   <2e-16 ***
kmercov 1.996e+01  2.452e-02  814.12   <2e-16 ***
bias    1.900e+00  2.495e-02   76.17   <2e-16 ***
length  4.553e+08  8.993e+05  506.34   <2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 4024000 on 1995 degrees of freedom

Number of iterations to convergence: 8 
Achieved convergence tolerance: 1.49e-08

Model for kmer plot on the right

Formula: y_transform ~ x^transform_exp * length * predict2_0(r1, k, d, 
    kmercov, bias, x)

Parameters:
         Estimate Std. Error t value Pr(>|t|)    
d       4.399e-02  1.407e-03   31.27   <2e-16 ***
r1      1.519e-02  1.002e-04  151.64   <2e-16 ***
kmercov 2.069e+01  2.416e-02  856.53   <2e-16 ***
bias    1.697e+00  2.099e-02   80.88   <2e-16 ***
length  5.069e+08  1.056e+06  480.10   <2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 4878000 on 1995 degrees of freedom

Number of iterations to convergence: 7 
Achieved convergence tolerance: 1.49e-08

Both the GenomeScope estimated genome sizes are smaller than the assembly (both have microorganisms filtered). The genome size of the left species is ~100Mb smaller than the one on the right, consistently in GenomeScope and the assembly.

GenomeScope Plots:

Answer 1

I would say yes, but it is not that easy to see it on a histogram decomposed to overlapping distributions. While the decomposition to different ditributions is assembly dependent, the overall k-mer histogram is actually directly derived from the sequencing data.

And these k-mer histograms can be used to estimate heterozygosity. For example, using genomescope. You can install the tool locally, or fit the model on a webserver (which seems to be down at the moment, but it should be back on at some point). For more details check the paper. If you get the tool running, you can estimate heterozygosity in the two samples to see how much more/less heterozygous they are.

P.S. I am always thinking about heterozygosity, which is supposed an equivalent of 1 - homozygosity, so I hope the answer will make sense to you.

--- edit ---

so, looking at the gneomescope models and knowing that genome size estimate matches the assembly size, I think it's fair to conclude the right genome is more heterozygous than the left one. However, keep in mind that it is not super great sequencing dataset, as the error peak and heterozygous peak are not clearly separated, hence the estimates of genomic features might be less accurate.