Genotype frequency dynamics in finite-sized, partially clonal population with mutation

This paper presents a Wright-Fisher-like model demonstrating that while partial clonality in finite populations does not alter mean allele frequencies or equilibrium states, it uniquely shapes the temporal dynamics of genotype frequencies and variance, thereby explaining observed deviations from Hardy-Weinberg equilibrium and enabling the inference of clonal rates from time-series genetic data.

The Gist

The Big Picture: The "Half-Sleeping" Population

Imagine a population of organisms (like plants, fungi, or certain animals) that has a split personality. Sometimes, they reproduce sexually (mixing genes from two parents, like shuffling two decks of cards). Other times, they clone themselves (making an exact photocopy of one parent).

Most scientists know how to predict what happens in a population that only shuffles cards (sexual) or a population that only photocopies (clonal). But figuring out what happens in a population that does both at the same time has been a mathematical nightmare. It's like trying to predict the weather when you have two different climate models that don't quite fit together.

This paper builds a new "weather model" for these half-cloning, half-sexual populations. It tracks how their genetic makeup changes over time, even when the population is small and random mutations (typos in the genetic code) happen.

---

The Core Discovery: The Two-Step Dance

The authors discovered that no matter how these populations start out, their genetic journey always follows a specific two-step dance:

Step 1: The "Straightening Out" Phase

Imagine a crowd of people standing in a chaotic, messy pile.

• Sexual populations straighten out into an orderly line almost instantly (in one generation).
• Clonal populations are stubborn. If they start messy, they stay messy for a long time because they are just photocopying the mess.
• The Finding: In a mixed population, the "clonality" acts like a brake. The more cloning happens, the slower the population gets back to a "balanced" state (called Hardy-Weinberg proportions). It's like a car with a heavy foot on the brake; it takes longer to reach the smooth highway.

Step 2: The "Slide Down the Hill" Phase

Once the population is balanced (Step 1 is done), it enters Step 2.

• Imagine a smooth, curved slide (the "Hardy-Weinberg parabola").
• Once the population is on this slide, it starts sliding toward a specific destination at the bottom (the Stable Equilibrium).
• The Surprise: It doesn't matter if the population is 90% clonal or 10% clonal. Once they are on the slide, they all slide at the same speed. The speed is determined only by how fast mutations (typos) happen, not by how much they clone.

The Analogy: Think of a river.

1. Phase 1: If you throw a leaf into a turbulent, rocky part of the river, a fast current (sexual reproduction) pushes it to the smooth water immediately. A slow current (clonality) lets it bounce around the rocks for a long time.
2. Phase 2: Once the leaf hits the smooth, fast-moving channel, it doesn't matter how it got there. It flows downstream at the speed of the water (mutation rate) toward the ocean (equilibrium).

---

Why Does This Matter? (The "Fis" Mystery)

In genetics, scientists use a number called $F_{IS}$ to measure if a population has too many identical twins (homozygotes) or too many mixed couples (heterozygotes).

• Positive $F_{IS}$: Too many clones/identical pairs.
• Negative $F_{IS}$: Too many mixed pairs (heterozygotes).

The Old Confusion: Scientists were confused because partially clonal populations sometimes showed positive numbers and sometimes negative numbers. It seemed random.

The New Explanation:

• During the "Messy Phase" (Step 1): If a population starts with too many clones, it will show a positive number. If it starts with too many mixed pairs, it shows a negative number. The "clonality rate" just changes how long it takes to fix this mess.
• During the "Slide Phase" (Step 2): Once the population is balanced, the number hovers near zero.
• The Twist: Because the population is finite (not infinite), random chance (genetic drift) makes the population wobble slightly around the perfect line. Interestingly, the shape of this wobble makes it slightly more likely to see a tiny bit of "too many mixed pairs" (negative $F_{IS}$) just by chance, especially in small populations.

The Takeaway: If you see a population with a mix of positive and negative numbers, or a high variance in these numbers across different genes, it's a strong signal that the population is partially clonal.

---

The "Concentration Ellipse" (The Fog of Uncertainty)

Since the population is finite, we can't predict the exact future; we can only predict the average future and how much it might wiggle around that average.

• The authors use a shape called a Concentration Ellipse to visualize this. Think of it as a "fog" surrounding the average path.
• The Big Discovery: The size and shape of this "fog" do not depend on how much cloning is happening.
• The fog is only determined by:
  1. How big the population is (smaller population = bigger fog).
  2. What the current genetic mix looks like.

This means that even if you don't know the cloning rate, you can still predict how much the genetics will "jitter" around the average path.

---

Why Should You Care?

1. Better Monitoring: Conservationists and ecologists often track populations over time (e.g., checking a forest every year). This new model allows them to look at genetic data from year to year and accurately figure out: "Is this population mostly cloning itself, or mostly mixing genes?"
2. Invasive Species & Pathogens: Many pests and diseases (like malaria or invasive weeds) use partial clonality. Understanding their genetic "dance" helps us predict how fast they will adapt to new drugs or environments.
3. Solving the Math Puzzle: The authors created a new mathematical tool (a Python library called DYGENCLON) that makes these complex calculations fast and easy, replacing old methods that were too slow or impossible for large populations.

Summary in One Sentence

This paper reveals that partially clonal populations follow a predictable two-step genetic journey: first, they slowly straighten out from their starting chaos (slowed down by cloning), and then they slide smoothly toward a stable future (driven only by mutation), allowing scientists to finally decode the hidden "cloning rate" of nature's most versatile species.

Technical Summary

1. Problem Statement

Most eukaryotes reproduce via partial clonality (a mix of sexual and asexual reproduction), yet appropriate population genetic models for these systems remain limited. Existing models often fail to accurately reconstruct past dynamics, predict future trajectories, or infer evolutionary processes in these populations.

• The Challenge: In fully sexual populations, genetic evolution can often be approximated by tracking allele frequencies. However, in partially clonal populations, genotypes must be tracked explicitly because clonal reproduction preserves specific genotype combinations.
• Computational Barrier: Traditional Markov chain models that track the probability of every possible genotypic state become computationally intractable as population size ($N$) and the number of alleles ($k$) increase. The number of possible states ($E_g$) grows factorially, making exact calculations impossible for realistic population sizes (e.g., $N=100$ with 4 alleles results in $>4 \times 10^{12}$ states).
• The Gap: There is a need for a model that can track the mean and variance of genotype frequencies over time in finite, mutating, partially clonal populations without requiring the enumeration of every possible genotypic state.

2. Methodology

The authors developed a Wright-Fisher-like model tailored for finite-sized, partially clonal populations with mutation.

• Model Framework:

  • Population: Finite size ($N$), diploid, non-overlapping generations.
  • Reproduction: A rate of clonality ($c$) defines the proportion of offspring produced asexually (mitosis) vs. sexually (random mating).
  • Mutation: A K-allele mutation (KAM) model where alleles mutate at a fixed rate $u$ with reciprocal and equal probabilities between all pairs.
  • State Space: Instead of discrete states, the model uses a continuous distribution of genotype frequencies. The system is treated as a Markov chain where the state is the vector of genotype frequencies.

• Mathematical Formulation:

  • The model derives equations for the genotype frequencies in the next generation ($n+1$) based on the current generation ($n$), separating contributions from clonal ($p^\#$) and sexual ($p^s$) pools.
  • Clonal Pool: Genotypes are copied with potential mutation.
  • Sexual Pool: Genotypes are formed by random union of gametes (Hardy-Weinberg proportions) with mutation.
  • Stochasticity: The number of individuals of each genotype follows a multinomial distribution $\mathcal{M}(N, \mathbf{p}_{n+1})$. For large $N$, this is approximated by a multinormal distribution.
  • Visualization: To handle the 3D simplex (for a 2-allele system: $aa, aA, AA$), the authors transformed coordinates into a 2D subspace ($\Pi^*$) to visualize trajectories and concentration ellipses (representing variance/dispersion).

• Tools:

  • The authors developed a Python/Numpy library named DYGENCLON to compute mean trajectories, variances, and generate visualizations.

3. Key Contributions

1. Continuous Approximation: The paper shifts from discrete state-space enumeration to a continuous density function approach, enabling the analysis of large populations where exact Markov chain calculations are impossible.
2. Two-Phase Dynamic Characterization: The model identifies a universal two-phase dynamic for populations out of equilibrium:
   • Phase 1: Rapid (or slowed by clonality) convergence toward the Hardy-Weinberg (HW) parabola.
   • Phase 2: Iteration along the HW parabola toward a stable equilibrium.
3. Decoupling of Mean and Variance: The study demonstrates that while clonality affects the speed and path of the mean trajectory, it does not affect the variance (dispersion) around that mean. Variance depends only on parental genotype frequencies and population size.
4. Invertibility: The mean trajectories are shown to be bijective (one-to-one and invertible), suggesting that past and future genotype frequencies can be inferred from time-series data.

4. Key Results

A. Dynamics of Genotype Frequencies

• Universal Equilibrium: Regardless of the rate of clonality ($c$), all populations converge to the same unique stable equilibrium determined solely by reciprocal mutation rates (e.g., for 2 alleles with equal mutation, the equilibrium is $p=0.25, q=0.5, r=0.25$).
• The Two-Phase Process:
  1. Return to HW: Populations not in HW proportions move toward the HW parabola. Higher clonality rates slow down this return.
  2. Iteration along HW: Once on the parabola, populations move toward the equilibrium. The path along the parabola is identical for sexual populations but slightly deviated for highly clonal populations due to the time lag in the first phase allowing mutation to shift allele frequencies.
• Effect of Clonality on Means: Clonality changes the speed and direction of the mean trajectory out of the HW parabola but does not alter mean allele frequencies or gene diversity (heterozygosity) along the trajectory.

B. Stochasticity and Concentration Ellipses

• Independence from Clonality: The size and shape of the concentration ellipses (representing the variance of genotype frequencies) depend only on the current genotype frequencies and population size ($N$). They are independent of the clonality rate.
• Ellipse Behavior: The area of the ellipse is maximal in the interior of the simplex and diminishes near boundaries. Near the equilibrium, the ellipse exhibits eccentricity, favoring slight heterozygote excesses due to the concavity of the HW parabola.

C. Implications for $F_{IS}$ (Inbreeding Coefficient)

• Transient $F_{IS}$: Positive or negative $F_{IS}$ values are expected primarily during the transient phase when the population is returning to the HW parabola.
• Negative $F_{IS}$ (Heterozygote Excess): In finite populations, even at equilibrium, a slight excess of negative $F_{IS}$ (heterozygote excess) is expected. This is due to the geometric properties of the concentration ellipse relative to the concave HW parabola.
• Clonality Inference:
  • The variance of $F_{IS}$ across loci is a robust proxy for inferring clonality rates.
  • The number of generations required for a population to return to HW proportions after a perturbation increases with the rate of clonality. This temporal signature allows for the inference of clonality rates from time-series monitoring data.

5. Significance and Applications

• Monitoring and Management: The model provides a theoretical framework for analyzing time-series genotyping data from monitored populations (e.g., pathogens, invasive species, conservation targets). It allows researchers to predict future genetic trajectories and reconstruct past dynamics.
• Inference of Reproductive Modes: The findings suggest that the rate of clonality can be inferred by observing how long it takes for genotype frequencies to return to Hardy-Weinberg proportions or by analyzing the variance of $F_{IS}$ across loci.
• Overcoming Computational Limits: By using a continuous approximation, the model makes it feasible to study genetic drift and mutation in large, partially clonal populations where exact discrete models fail.
• Future Directions: The authors note that the model can be extended to non-reciprocal mutation rates (common in SNPs) and variable population sizes, paving the way for more realistic applications in evolutionary biology and epidemiology.

In summary, this paper bridges a critical gap in population genetics by providing a tractable, stochastic model for partially clonal populations, revealing that while clonality alters the tempo of evolution, the destination (equilibrium) and the noise (variance) follow predictable rules distinct from the mean trajectory's path.