Effects of selection and mutation on epidemiology of X-linked genetic diseases

  • Received: 02 March 2016 Accepted: 11 November 2016 Published: 01 June 2017
  • MSC : Primary: 58F15, 58F17; Secondary: 53C35

  • The epidemiology of X-linked recessive diseases, a class of genetic disorders, is modeled with a discrete-time, structured, non linear mathematical system. The model accounts for both de novo mutations (i.e., affected sibling born to unaffected parents) and selection (i.e., distinct fitness rates depending on individual's health conditions). Assuming that the population is constant over generations and relying on Lyapunov theory we found the domain of attraction of model's equilibrium point and studied the convergence properties of the degenerate equilibrium where only affected individuals survive. Examples of applications of the proposed model to two among the most common X-linked recessive diseases (namely the red and green color blindness and the Hemophilia A) are described.

    Citation: Francesca Verrilli, Hamed Kebriaei, Luigi Glielmo, Martin Corless, Carmen Del Vecchio. Effects of selection and mutation on epidemiology of X-linked genetic diseases[J]. Mathematical Biosciences and Engineering, 2017, 14(3): 755-775. doi: 10.3934/mbe.2017042

    Related Papers:

    [1] Wei Yin, Tao Yang, GuangYu Wan, Xiong Zhou . Identification of image genetic biomarkers of Alzheimer's disease by orthogonal structured sparse canonical correlation analysis based on a diagnostic information fusion. Mathematical Biosciences and Engineering, 2023, 20(9): 16648-16662. doi: 10.3934/mbe.2023741
    [2] Beryl Musundi . An immuno-epidemiological model linking between-host and within-host dynamics of cholera. Mathematical Biosciences and Engineering, 2023, 20(9): 16015-16032. doi: 10.3934/mbe.2023714
    [3] Xue Liu, Xin You Meng . Dynamics of Bacterial white spot disease spreads in Litopenaeus Vannamei with time-varying delay. Mathematical Biosciences and Engineering, 2023, 20(12): 20748-20769. doi: 10.3934/mbe.2023918
    [4] Luju Liu, Jianhong Wu, Xiao-Qiang Zhao . The impact of migrant workers on the tuberculosis transmission: General models and a case study for China. Mathematical Biosciences and Engineering, 2012, 9(4): 785-807. doi: 10.3934/mbe.2012.9.785
    [5] Süleyman Cengizci, Aslıhan Dursun Cengizci, Ömür Uğur . A mathematical model for human-to-human transmission of COVID-19: a case study for Turkey's data. Mathematical Biosciences and Engineering, 2021, 18(6): 9787-9805. doi: 10.3934/mbe.2021480
    [6] Jianjun Tian, Bai-Lian Li . Coalgebraic Structure of Genetic Inheritance. Mathematical Biosciences and Engineering, 2004, 1(2): 243-266. doi: 10.3934/mbe.2004.1.243
    [7] Qianqian Li, Ankur Jyoti Kashyap, Qun Zhu, Fengde Chen . Dynamical behaviours of discrete amensalism system with fear effects on first species. Mathematical Biosciences and Engineering, 2024, 21(1): 832-860. doi: 10.3934/mbe.2024035
    [8] Yongli Cai, Yun Kang, Weiming Wang . Global stability of the steady states of an epidemic model incorporating intervention strategies. Mathematical Biosciences and Engineering, 2017, 14(5&6): 1071-1089. doi: 10.3934/mbe.2017056
    [9] Ilse Domínguez-Alemán, Itzel Domínguez-Alemán, Juan Carlos Hernández-Gómez, Francisco J. Ariza-Hernández . A predator-prey fractional model with disease in the prey species. Mathematical Biosciences and Engineering, 2024, 21(3): 3713-3741. doi: 10.3934/mbe.2024164
    [10] Yuqing Qian, Tingting Shang, Fei Guo, Chunliang Wang, Zhiming Cui, Yijie Ding, Hongjie Wu . Identification of DNA-binding protein based multiple kernel model. Mathematical Biosciences and Engineering, 2023, 20(7): 13149-13170. doi: 10.3934/mbe.2023586
  • The epidemiology of X-linked recessive diseases, a class of genetic disorders, is modeled with a discrete-time, structured, non linear mathematical system. The model accounts for both de novo mutations (i.e., affected sibling born to unaffected parents) and selection (i.e., distinct fitness rates depending on individual's health conditions). Assuming that the population is constant over generations and relying on Lyapunov theory we found the domain of attraction of model's equilibrium point and studied the convergence properties of the degenerate equilibrium where only affected individuals survive. Examples of applications of the proposed model to two among the most common X-linked recessive diseases (namely the red and green color blindness and the Hemophilia A) are described.


    1. Introduction

    Epidemiological studies concern with models of the occurrence of diseases in space and time. They are carried out relying on mathematical models and their analysis' tools; mathematic aids in inferring disease causes, predicting the future course of an outbreak and planning the most appropriate control measures. The most numerous epidemiological mathematical models are in the field of infectious diseases [11,19,24]. Under different assumptions, dynamical systems have been designed to predict the number of infected people over times, as well as to quantify the effect of diseases' spread control measures such as pharmacological therapies or social actions. [2,22,41].

    When referring to genetic diseases, epidemiological studies examine the role of genetic factors and their interaction with environmental features in the occurrence of human genetic diseases i.e., an illness caused by abnormalities within the DNA sequence of human genes. Their general aim is to understand the role of genetic factors in the etiology of diseases in human populations with the ultimate goal of disease control and prevention [25].

    Mathematical models and analyses have successfully been applied in human genetics epidemiology (see [25,26,27] and reference therein). Dynamic models have been developed to describe gene distributions in evolving populations, as well as changes in a single genome or biochemical process regulated by genes [16,43,17]. Mathematical models can help to provide non intuitive insights into disease spread within a population as well as to predict the influence of de novo gene mutations on the incidence of these diseases or quantify the effectiveness of treatments [7,3,44].

    We studied the epidemiology of a class of genetic disorders, namely the X-linked recessive diseases; these include the serious diseases hemophilia A, Duchenne/Becker muscular dystrophy, and Lesch-Nyhan syndrome as well as common and less serious conditions such as male pattern baldness and red-green colour blindness. Although genetic diseases are very rare, it has been estimated that millions of people are affected worldwide [21]. The incidence of an X-linked recessive disease depends on the disorder's severity: it ranges from 1 in 3000 newborn males for Duchanne/Becker muscular dystrophy to 1 in 20 for the red and green color blindness.

    Despite their relevance, only a few mathematical models have been specifically developed to study the dynamics and spread of X-linked recessive diseases in a population [43,14]. Related studies analyze the inheritance mechanism of any gene-not necessarily responsible for a genetic disease-placed on the X chromosome [29,28]; they belong to the field of population genetics. In these works genotypes frequencies-i.e., the frequency or proportion of genotypes in a population-are often chosen as model's variables. Under the hypothesis of infinite population and starting from a genotypes' distribution, the genotypes' proportions in the next generation are evaluated according to the inheritance mechanism and to the effects of selection or mutation. The average fitness (see [30] pag. 385-387) is frequently studied as a suitable Lyapunov function candidate to analyze stability properties of model's equilibrium points. Even in this generic scenario seldom contributions examine the combined effects of selection and mutation on population's dynamics and equilibrium (see [31,39] and reference therein). Furthermore findings of these researches can not be applied to epidemiological studies. In fact the ultimate goal of genetic epidemiology is to predict the number of individuals carrying the disease responsible gene; this number can not be inferred from genotypes' frequency distribution when assuming infinite population size.

    We developed a discrete time mathematical model aiming at describing the spread of X-linked recessive diseases in the population as well as quantifying the influence on diseases distribution of related factors such as individual's fitness and sporadic genetic mutation.

    In our preliminary studies [14,15] we developed an X-linked recessive diseases model which did not take into account the class of affected women. In [14] we made the simplifying hypothesis that only sons born from healthy couples could be born affected owing to spontaneous mutations. Fitness rates were assumed identical for all individuals, regardless of their health conditions. In that simplified scenario we proved that the population asymptotically distributes among classes in proportions depending on the spontaneous mutation rate values. In [15] we studied the influence of fitness factors on the same model framework and we inferred general system properties assuming that either selection or mutation apply.

    Encouraged by previous results and leveraging on the same framework, we elaborated the model we are here presenting which includes the class of affected women leading to a five state system. We also took into account genetic mutations possibly occurring to any couples' offspring and fitness factors varying according to individual's health conditions. These features enable to reproduce the inheritance mechanism and the spread in a population of any X-linked disease.

    Although the mathematical model we developed is nonlinear, it is suitable to certain analyses using classical nonlinear methods to gain information about system behavior, equilibrium existence and their convergence properties. We remark that in genetic epidemiology convergence means that a population initialized at some point (i.e., with an assigned initial allocation among healthy, affected and carrier people) will move, through a time sequence of generations, to a (possibly) different population distribution [10]. Similar analysis have been conducted in general biological systems: in [4] a Lyapunov method is proposed for finding the invariant sets of non-negative dynamical systems modeling overpopulation species in ecology, while in [8] equivalent results are obtained using Jacobian analysis.

    The organization of the paper is as follow. In Section 2 we introduce a background on X-linked recessive diseases, details on spontaneous genetic mutation and individual's fitness factors. After giving details on the assumptions and on the inheritance mechanisms of X-linked diseases we propose a mathematical model in Section 3 and its equilibrium points are analyzed in Section 4. Simulation results and a sensitivity analysis are given respectively in Section 5 and 6. Finally, some discussions and conclusions are presented in Section 7 and 8.


    2. The X-linked recessive inheritance mechanism

    Clinical expression and inheritance patterns of X-linked recessive diseases are related to individual gender. While females possess two X chromosomes, males have one X chromosome and one Y chromosome; thus females possess two copies of each X chromosome gene, whereas males only have a single copy of each X and Y chromosome genes [16].

    X-linked diseases can be either recessive or dominant [18]. In X-linked recessive diseases females harbor one copy of each the disease responsible and the normal gene. As the effects of the abnormal gene are counteracted by those of the normal gene located on the alternative X chromosome, females usually show no signs of disease. For that reason, they are called carriers, and can transmit either the normal or the abnormal gene to their progeny. Thus, to become affected by an X-linked recessive disease, females have to harbor a copy of the disease-related gene on both X chromosomes. In contrast, males harboring an abnormal gene on their unique X chromosome are usually affected, even if the disease is recessive. Indeed, the Y chromosome lacks all X chromosome genes, and is therefore unable to exert any compensatory effect.

    An X-linked recessive disease may be inherited according to the following rules:

    • Affected males never spread the disease to their sons, as no male-to-male transmission of the X chromosome occurs (see Figure 1(a)).

    Figure 1. Inheritance pattern of X-linked recessive disease.

    • Affected males pass the abnormal X chromosome to all of their daughters, who are described as obligate carriers (see Figure 1(a)).

    • On average, female carriers pass a defective X chromosome to half of their sons (who will born affected) and half of their daughters (who become carriers). The remaining half of their siblings inherit a normal copy of the chromosome (see Figure 1(b)).

    • Affected females are the rare result of an affected male and a carrier or affected female mating. They are more frequent in less serious X-linked recessive diseases that do not comprise reproduction capacities.

    We refer the reader to [18] for details on clinical expression and inheritance patterns of X-linked recessive diseases.

    Other than the result of the described, well characterized patterns of inheritance, the current spread of genetic disorders within a given population is influenced by additional factors, including sporadic mutations and individual's fitness. Both are "driving forces" in the transmission mechanism of X-linked recessive diseases and they will be described in the next section; other factors such as prenatal diagnosis, population migrations and non random mating slightly affect diseases' spread [31,35].


    2.1. Spontaneous genetic mutation

    A genetic disease that occurs when neither parent is affected or carrier of any genetic defect is called sporadic mutation or de novo gene mutations. These cases arise via random genetic mutations within the DNA sequence; the mutation can occur in the germ-line cell population-i.e. in eggs and sperm cells-in subjects without any prior genetic defect and it can be transmitted down to one of the offspring [5]. The genetic mutation can also occur in the zygote cell, i.e. the initial cell formed when two gametes join.

    A sporadic mutation can be the cause of an X-linked recessive disease (whereas it is unlikely for an autosomal recessive disorder) as a single mutation is enough in males to cause the disease. Males can be born affected due to a spontaneous gene mutation as a single abnormal gene copy is enough for the disease to become manifest; females can also be born carriers owing to random mutations.

    Gametes are haployd cells, i.e., they only have one copy of the sex chromosomes (X or Y). To model the effects of spontaneous mutation on the diseases' distribution among the population, it is necessary to quantify the probability of inheriting an abnormal gene copy due to spontaneous mutation. Mutation may occur in any gene of the X chromosome, not all of them will cause a disease. From here on we consider only those mutations that occur on disease related genes. If the effects of de novo mutations are not considered, a carrier female has half of her gametes carrying a defective gene copy (Xa) and half hosting a normal gene (XA) while the gametes of a healthy female contains only normal gene copies. All X gametes of an affected male carry the abnormal gene copy.

    We let γ denote the probability of a relevant mutation in the X chromosone of a gamete changing a normal gene copy into an abnormal one. Owing to spontaneous mutation, a healthy male has a probability equal to γ of transmitting a defective gene copy to his daughters; consequently, 1γ is the probability of transmitting a normal gene copy. In contrast, all the gametes of an affected male host only defective X chromosomes; hence the daughter of an affected male will receive a defective X chromosome with probability one.

    Similarly, a healthy woman has a probability equal to γ or 1γ to transmit a defective gene copy or a normal gene copy to any of her progeny respectively. The combinations of these probabilities merged with the female and male X-linked inheritance pattern lead to the probabilities of having a healthy, carrier or affected child reported in Tables 1 and 2. Specialized medical literature, (see for example [12,37]) report γ ranges in [0,105]; the rate of de novo mutations varies widely among different geographic regions, and depends on a number of factors, including environmental exposure to mutagenic agents, breadth of gene sequence and ability of the cell machinery to actually repair or correct the mutations [13].

    Table 1. X-linked recessive inheritance probabilities for sons.
    PARENTS SONS
    father mother healthy
    XAY
    affected
    XaY
    XAY XAXA 1γ γ
    XAY XAXa 12(1γ) 12(1+γ)
    XAY XaXa 0 1
    XaY XAXA 1γ γ
    XaY XAXa 12(1γ) 12(1+γ)
    XaY XaXa 0 1
     | Show Table
    DownLoad: CSV
    Table 2. X-linked recessive inheritance probabilities for daughters.
    PARENTS DAUGHTERS
    father mother healthy
    XAXA
    carrier
    XAXa
    affected
    XaXa
    XAY XAXA (1γ)2 2γ(1γ) γ2
    XAY XAXa 12(1γ)2 12(1+γ2γ2) 12γ(1+γ)
    XAY XaXa 0 1γ γ
    XaY XAXA 0 1γ γ
    XaY XAXa 0 12(1γ) 12(1+γ)
    XaY XaXa 0 0 1
     | Show Table
    DownLoad: CSV

    Recent studies have shown that the longer is the disease's responsible gene the greater is the probability of spontaneous mutations.

    Finally we notice that some recent studies suggest different mutation rates between females and males. This topic is currently discussed in the scientific community and there are not shared opinions (see for instance [42] [39]); the above differentiation is beyond the purpose of our work which specifically refers to diseases' responsible gene; hence in what follows we will adopt the mutation rate γ to both females and males changing the non defective gene A into a.


    2.2. Individual fitness

    The contribution of an individual to the next generation varies according to the individual's health and it is usually referred as individual's fitness [30,33]. Affected individuals may be in various ways disadvantaged to reproduce compared with healthy ones: this might be due to a lower likelihood to survive, hence to reach reproduction age, or to reduced fertility. These individuals, accordingly, will have relatively less children compared with healthy ones thus reducing the probability of passing the disease responsible gene to the next generation. This selective difference mechanism leads to a gradual evolution in the population's proportions among healthy, affected and carriers.

    Following [32], the fitness factors wi are the product of the viability, i.e., the probability that an individual survives until the reproduction age, and the fertility, i.e., the average number of individual progeny. Hence the fitness factor wi-also called fertility factor-models reproduction capacity of individuals of class i according to their health conditions. The parameters wi are positive and are bounded by clinical considerations: the more severe the disease the smaller are their values in affected individuals.


    3. Mathematical model of X-linked recessive diseases

    In an attempt to further improve the predictive value of genetic disease' spread within a population, we designed a detailed model of the epidemiology of X-linked recessive diseases. Our model fits in the category of structured models [16,17]. The studied population is divided into homogeneous groups in relation to some major parameters (such as subject's age, sex or health conditions).

    We divide the population into five classes, according to sex and health conditions (namely healthy and affected men, healthy, carrier and affected women) and, unlike other work in the literature, we choose the number of individuals in each class as system variables. Thus we introduce variables x1 and x2 to denote the number of healthy and affected males respectively and x3, x4 and x5 to denote the number of healthy, carrier and affected females. This allows to reproduce the population distribution in the five classes at each generation and to predict the number of healthy, affected and carrier individuals at time.

    We make the following assumptions:

    • time is discrete and denotes generations;

    • generations are non-overlapping thus each person breeds with a person of the same generation;

    • the population is constant and equal to 2N, that is,

    x1(k)+x2(k)+x3(k)+x4(k)+x5(k)=2Nk; (1)

    • in each generation there is an equal number of males and females, that is,

    x1(k)+x2(k)=x3(k)+x4(k)+x5(k)k. (2)

    • The number of sons (equal to number of daughters) of each couple varies according to parents' health conditions and this is modeled through the fitness factors wi for i=15 that will be illustrated below.

    • Spontaneous genetic mutations are modeled according to their probability of occurrence described in Section 2.1.

    The number of males born from a person of class i{1,2} with a person of class j{3,4,5} is:

    12wiwjxi(x1+x2)xj(x3+x4+x5)(x1+x2+x3+x4+x5)=wiwjxixjN (3)

    where wi is the fitness factor of a person of class i.

    For example consider a couple formed by a healthy father (an individual of class 1) and a healthy mother (an individual of class 3); γ sons of this couple (on average) will be affected and ˜γ will be healthy. Consider all couples of class 1,3 in the population; the number of their healthy sons at the next generation will increase of

    ˜γ2w13x1(x1+x2)x3(x3+x4+x5)(x1+x2+x3+x4+x5). (4)

    In view of the assumptions (1) and (2) system dynamics can be written using one state variable of the male population (i.e., x1 or x2) and two variables of the female population (i.e., two among x3, x4 or x5). In the following we will restrict our analysis to healthy males and healthy and carrier females dynamics.

    Let x=[x1x3x4]T be the state variables vector. The population dynamics are described through the following nonlinear, discrete-time system:

    x+=f(x) (5)

    where x+(k):=x(k+1), f=[f1f3f4]T and

    f1(x)=1N[w13˜γx1x3+12w14˜γx1x4+w23˜γ(Nx1)x3+12w24˜γ(Nx1)x4] (6a)
    f3(x)=1N[w13˜γ2x1x3+12w14˜γ2x1x4] (6b)
    f4(x)=1N[2w13γ˜γx1x3+w23˜γ(Nx1)x3+w15˜γx1(Nx3x4)+12w24˜γ(Nx1)x4+12w14(1+γ2γ2)x1x4] (6c)

    with

    ˜γ:=1γandwij:=wiwj.

    Term (4) is the first element in the right side of equation (6a); all other terms are derived with similar reasoning.

    System (5)-(6) has been designed according to the inheritance pattern of X-linked recessive disease described in Section 2, the spontaneous mutation occurrence in Section 2.1, and given the newborn children as in (3).

    We explicitly note that negative solutions of the dynamic equations (5)-(6) are not meaningful for an epidemiological model. The positivity of the state trajectories of system (5)-(6) is guaranteed assuming wij within appropriate sets that may be derived exploring systems equation; an example of how these sets may be inferred is reported in the Appendix 8.

    For the analysis in the following sections, it is helpful to rewrite system (5)-(6) in the form reported below:

    x+=Ax+x1Bx (7)

    with

    A=[0a12a13000a31a32a33],B=[0b12b130b22b230b32b33] (8)

    where:

    a12=a32=w23˜γa13=a33=12w24˜γa31=w15˜γb12=1N˜γ(w13w23)b13=12N˜γ(w14w24)b22=1Nw13˜γ2b23=12Nw14˜γ2b32=1N[2w13γ˜γw23˜γw15˜γ]b33=1N[12w14(1+γ2γ2)w15˜γ12w24˜γ].

    4. System's equilibrium analysis

    System (7) may admit many equilibrium points; according to [38] less than 23. One is the zero vector ˉx=(0,0,0) which implies x2=x5=N; when a population reaches this equilibrium, individuals are all affected.

    Other equilibrium states depend on wi,i=1,,5, and γ; if one component of the equilibrium point is either negative or greater than N, then it is unfeasible due to model's hypothesis.

    Let ˉx be a generic equilibrium point for system (7) and define yxˉx. We rewrite (7) as

    y++ˉx=A(y+ˉx)+(y1+ˉx1)B(y+ˉx).

    Since ˉx=Aˉx+ˉx1Bˉx we obtain

    y+=˜Ay+y1By (10)

    where

    ˜A=A+[ˉx3b12+ˉx4b13ˉx1b12ˉx1b13ˉx3b22+ˉx4b23ˉx1b22ˉx1b23ˉx3b32+ˉx4b33ˉx1b32ˉx1b33].

    Note that convergence of system (10) to ˉy=(0,0,0) corresponds to convergence of system (7) to ˉx.

    Theorem 4.1. If matrix ˜A is Schur then, the equilibrium state ˉx of system (10) is asymptotically stable with

    Λ={x|(xˉx)TH(xˉx)<r} (11)

    as a region of attraction where

    r=δ2/λ4δ:=(bλ1)2+λ1λ3bλ1λ1:=λmin(H1G)λ2:=1/λmin(H)λ3:=λmax(H1BTHB)λ4:=λmax(H1CCT)b:=λ2BTH˜AC=[100]T

    and G and H are positive definite symmetric matrices satisfying

    ˜ATH˜AH=G. (12)

    The proof is carried out considering the following Lyapunov function

    V(y)=yTHywithy=xˉx. (13)

    Detailed demonstration is reported in the Appendix 8.

    Remark 1. Theorem 4.1 for equilibrium point ˉx=(0,0,0) requires that matrix A is Schur; ranges of the fertility factors that guarantee this condition are given in the Appendix 8.


    4.1. Stability analysis of the origin

    Convergence of system (7) to the equilibrium point ˉx=(0,0,0) coincides with the epidemiological situation where all the individuals are affected. Since this is not desirable it is useful to know the domain of attraction of ˉx=(0,0,0) and avoid it. As matter of fact any point in this region represents a population distribution among healthy, affected and carriers individuals that leads, towards generations, to all affected people. In what follows our gaol is to delineate a wider domain of attraction for ˉx=(0,0,0) compared to Λ in Theorem 4.1 and explore global stability. To this aim, we propose a candidate Lyapunov function for the zero equilibrium point of system (7); we will show through simulation that this Lyapunov function leads to a larger domain of attraction compared to Λ.

    In addition, we give conditions for global asymptotic stability of ˉx=(0,0,0) in terms of fertility factors wij.

    Consider the following candidate Lyapunov function:

    V(x)=αx1+βx3+μx4 (14)

    with α, β and μ>0. It holds:

    ΔV=V(x+)V(x)=(μa31α)x1+[(α+μ)a12β]x3+[(α+μ)a13μ]x4+x1[(αb12+βb22+μb32)x3+(αb13+βb23+μb33)x4]. (15)

    4.1.1. Local stability analysis

    The sufficient conditions for local stability (i.e., on linear terms in (15)) are:

    μa31α<0w15<αμ˜γ (16a)
    (α+μ)a12β<0w23<β˜γ(α+μ) (16b)
    (α+μ)a13μ<0w24<2μ˜γ(α+μ). (16c)

    Let's assume that (16) hold; equation (15) can be rewritten as:

    ΔV=(μa31α)x1+[(α+μ)a12β+(αb12+βb22+μb32)x1]x3+[(α+μ)a13μ+(αb13+βb23+μb33)x1]x4

    in order to have ΔV<0, x1 should verify the following conditions:

    {x1<β(α+μ)a12αb12+βb22+μb32=k1k2x1<μ(α+μ)a13αb13+βb23+μb33=z1z2

    with k1>0 and z1>0. Depending on values of k2 and z2 we can derive the domain of attraction Φ of the equilibrium point (0,0,0):

    (ⅰ) k2>0, z2>0:

    Φ={x|x1<min(k1k2,z1z2,N)}; (19)

    (ⅱ) k2>0, z2<0: since z1>0 the domain of attraction is given by

    Φ={x|x1<min(k1k2,N)}; (20)

    (ⅲ) k2<0, z2>0: since k1>0 the domain of attraction is identified by

    Φ={x|x1<min(z1z2,N)}; (21)

    (ⅳ) k2<0, z2<0: the equilibrium point is globally asymptotically stable.

    Conditions on k2 and z2 in (i), (ii), (iii) or (iv) depends on w13 and w14 values; for example, holding (16), k2>0 and z2>0 if both of the following two conditions apply:

    w13>α+β˜γ(α+β˜γ+2μγ)w14>2(α+μ)α˜γ+β˜γ2+μ(1+γ2γ2).

    Proposition 1. Assume that wij verify (16) and one of conditions in (i), (ii) or (iii) holds, then the equilibrium point ˉx=(0,0,0) of system (7) is asymptotically stable within Φ.


    4.1.2. Global stability analysis

    Proposition 2. If wij are small enough such that constraints (16) hold and

    w13<(α+μ)w23+μw15α+β˜γ+2μγ (23a)
    w14<[(α+μ)w24+2μw15]˜γα˜γ+β˜γ2+μ(1+γ2γ2), (23b)

    then system (7) is globally asymptotically stable about (0,0,0).

    Proof. Constraints (23) on wij assure that k2<0, z2<0.

    Remark 2. So far we have introduced two Lyapunov functions; the first in (13) is suitable to analyze convergency properties of any equilibrium in system (10)-that is equivalent to system (5)-(6)-and gives a domain of attraction for the analyzed point. After that we designed the Lyapunov function (14) to specifically analyze the zero equilibrium point convergency properties. Compared to (13) the Lyapunov function (14) allows to:

    1. Obtain the sufficient conditions for system's global asymptotic stability referred to ˉx=(0,0,0).

    2. Avoid conservative relations, such as norm and eigenvalue inequalities, for determining the domain of attraction; this could imply that region Φ is larger than Λ although we were not able to formally prove this.

    3. Relax the conditions of local stability for two state variables, namely x3 and x4.


    4.2. Choosing optimal values of the parameters α, β and μ

    In this section we present a procedure to select the parameters' value α,β,μ of the Lyapunov function in (14). Due to the critical role of the zero equilibrium point, we choose these values such that the domain of attraction Φ of the equilibrium point (0, 0, 0) is maximized.

    The optimal values α, β and μ of the parameters are obtained by solving the following optimization problem:

    maxα,β,μΦ(x,α,β,μ)

    subject to

    μa31α<0(α+μ)a12β<0(α+μ)a13μ<0α,β,μ>0.

    The objective function to be maximized is one of the regions of attraction (19), (20) and (21) according to k2 and z2 signs; the constraints are given by the sufficient conditions (16) for the local stability of the zero equilibrium point.

    Genetic Algorithm (GA) method has been exploited for solving this constrained nonlinear optimization problem which is not well suited for standard optimization methods. GA peculiar mechanisms to avoid local minima make the use of a GA reasonable in facing problems with complex nonlinear objective function (like in our case) in place of gradient based methods which are designed to be fast and efficient for finding local minima.


    5. Simulations

    In what follows we present some simulations of system (7) that is equivalent to (5)-(6). We show results obtained in Section 4 on the local stability of the equilibrium point ˉx=(0,0,0); namely we evaluate the domains of attraction Λ and Φ obtained through the Lyapunov functions (13) and (14) respectively. Moreover we give results related to a non zero equilibrium and show its convergency according to Theorem 4.1. Finally we depict the trajectories for some of the analyzed scenarios.


    5.1. Domain of attraction for ˉx=(0,0,0).

    We assign three different sets of values to wi parameters, reported in Table 3; they have been chosen such that matrix A in system (7) is Schur (i.e., conditions (27) are satisfied).

    Table 3. Parameters values.
    N γ w13 w14 w15 w23 w24 r
    scenario 1 150 104 0.5 0.45 0.1 1 0.9 11232
    scenario 2 150 104 0.5 1 0.5 0.5 1 8284
    scenario 3 150 104 0.63 1.4 0.14 0.7 1.5 2121
     | Show Table
    DownLoad: CSV

    Using Theorem 4.1 we determine the domain of attraction Λ of the equilibrium point ˉx=(0,0,0) according to (11); Λ is an ellipsoid centered in the equilibrium point with semi-principal axes equal to the square roots of the reciprocals of matrix H eigenvalues multiplied by r. Matrix H is the solution of discrete Lyapunov equation (12) where matrix G has been chosen equal to the identity matrix. Figure 2 depicts the domains of attraction Λ for each scenario in Table 3, where the corresponding values of r are reported.

    Figure 2. Region of attraction corresponding to wi parameters in Table 3.

    To apply results in Proposition 1 we evaluate the coefficients of Lyapunov function (14) for each scenario in Table 3. To this aim we solve the optimization problem in Section 4.2 and we compute the domains of attraction Φ={x|x1<x_1}; the obtained α,β,μ and x_1 for the three scenarios are reported in Table 4.

    Table 4. Parameters values for Lyapunov function in (14).
    α β μ x_1
    scenario 1 0.0917 0.9999 0.9072 150
    scenario 2 0.2486 0.3729 0.4953 150
    scenario 3 0.0247 0.1359 0.1766 63.6
     | Show Table
    DownLoad: CSV

    Figure 3 depicts the domains of attraction Φ compared to the ellipsoids Λ. Note that in each of the reported scenarios Φ is sensibly greater than Λ; indeed for the first two scenarios Φ corresponds to system states space.

    Figure 3. Region of attraction corresponding to wi parameters in Table 3 using the two Lyapunov functions.

    5.2. Equilibrium point other than ˉx=(0,0,0).

    We consider w1=0.9, w2=2, w3=1, w4=1 and w5=0.5, γ=104 and N=150; for this choice of parameters system (5)-(6) has the following two equilibrium points: ˉxA=(0,0,0), ˉxB=(72.5,26.5,68.9). Note that these parameters do not satisfy (27); thus matrix A in system (7) is unstable, and system (5)-(6) will not converge to ˉxA. Finally matrix ˜A in system (10) evaluated at ˉxB is stable, and Theorem 4.1 applies. The corresponding domain of attraction Λ has r=341.45 and it is depicted in Figure 4.

    Figure 4. Region of attraction of xB = (72.5, 26.5, 68.9) and system's parameters as in Section 5.2.

    5.3. State portrait

    Figure 5 depicts trajectories of system (5)-(6) in state space with the wi parameters assigned according to the scenario 1 in Table 3. The initial conditions are chosen within the corresponding domain of attraction shown in Figure 2; as expected, all trajectories asymptotically go to zero.

    Figure 5. Trajectories with different initial conditions and wi in (27).

    Figure 6 represents trajectories of system (5)-(6) and convergency to ˉxB when initial conditions are inside the domain of attraction obtained in Section 5.2.

    Figure 6. Trajectories converging to xB.

    6. Sensitivity analysis

    Local sensitivities have been performed to analyze the dependency of the dimension of the domain of attraction Λ in (11) on system's fitness factors. Local sensitivity analysis is performed in a static way; the output obtained changing each parameter in a predefined interval and with constant step are evaluated. Namely we quantify the influence of couples' fitness factors wij on r in (11). The analysis is performed perturbing one parameter at time on M different values equally distributed on an interval. We set wij=ˉwij, and obtain the nominal trajectory rnom; rpert is obtained perturbing one parameter from its nominal value. For each perturbed parameter, we compute the following function:

    ΔEl=(rpertlrnom)2,

    where l=1,,M denotes the values of the considered perturbed parameter.

    To compute the nominal trajectory we consider γ=104, N=150 and we fix the nominal conditions [ˉw13,ˉw14,ˉw15,ˉw23,ˉw24]=[0.5,0.5,0.5,0.5,0.5]. The perturbed trajectories are obtained letting wij vary in the range [0,1.3], with a constant step of 0.01. Only one parameter varies per time; all others are set at their nominal value. To each combination of wij values corresponds a set of system's equilibrium points that always include ˉx=(0,0,0); in order to perform a meaningful comparison between the domains of attraction we only referred to the Λ of the zero equilibrium. The range in which the wij vary assures that the matrix A is always Schur.

    Figure 7 depicts the sensitivity functions with respect to each wij. A flat sensitivity curve represents a little influence of the parameter on the output; a steep curve implies a great influence. Function r in (11) appears to be most sensitive to w15 and then to w24, that are related to the sufficient conditions for the local stability of the system (see equation (27)).

    Figure 7. Static sensitivity analysis with respect to wi.

    7. Discussion

    Genetic epidemiological studies have the ultimate goal of estimating the spread of a disease in a population. This is quite a recent and rapidly expanding research filed, but the implications on individual or population health are still unclear [36]. Findings in genetic epidemiology are exploited to design genetic screening campaign aiming at assessing, through DNA sequencing tests, how many people are affected or carry the disease's responsible gene. Usually females carring a defective gene copy don't show any symptoms of the disease, thus their number is usually hard to estimate. State x4 of model (5)-(6) predicts the number of carrier in a population and may be successfully applied to genetic epidemiology.

    Genetic epidemiology may also contribute to measure the effect of environment's risk factors on disease spread. Urban areas exposed to toxins, radiations or other agents are periodically examined; sampled data on the spread of genetic diseases are compared with predicted values determined through mathematical models. By simulating model (5)-(6) one can evaluate the effects of fitness factors variations on spread of the disease. Simulations results may help in taking decisions on appropriate corrective strategies aiming at preventing the spread of the disease.

    Finally technological and other advances will allow the potential of genetic epidemiology to be revealed over the next few years, and the establishment of large population-based resources for such studies (biobanks) should contribute to this endeavor.

    Consider an X-linked disease in an assigned population 2N; parameters in model (5)-(6)-the mutation rate γ of the disease responsible gene and the fertility factors wii=1,,5-can be derived through clinical considerations or found in specialized medical literature. Applying the results on global asymptotic stability of equilibrium ˉx=(0,0,0), one can conclude that the population will asymptotically converge to all individuals affected if the assigned values of wi verify (16) and are within the range defined in (23). For the other equilibrium points of system (5)-(6) one can verify if matrix ˜A is stable and derive the corresponding region of attraction using Theorem 4.1 results. In what follows we apply our model and the results on stability analysis to two X-linked recessive diseases: red and green color blindness and Hemophilia A.


    7.1. Application 1

    Color blindness, or color vision deficiency, is the inability or decreased ability to see color, or perceive color differences, under normal lighting conditions. Color blindness affects a significant percentage of the population. The most common cause of the disease is a fault in the development of one or more sets of retinal cones that perceive color in light and transmit that information to the optic nerve. This type of color blindness is usually an X-linked condition as genes that produce photopigments are carried on the X chromosome; if some of these genes are missing or damaged, color blindness will be expressed in males with a higher probability because in females a functional gene on only one of the two X chromosomes is sufficient to yield the needed photopigments. According to statistical datas, [23] and [1], color blindness affects a significant number of people, although exact proportions vary among groups. Isolated communities with a restricted gene pool sometimes produce high proportions of color blindness, including the less usual types. Examples include rural Finland, Hungary, and some of the Scottish Isles. We apply our model to study the red and green color blindness diffusion in Scottish Isles. As reported in the National Records of Scotland, in 2011 Scottish Isles had a population of 87252. According to data in [34], the mutation rate for red and green color blindness (γ in our model), is approximately 2105. Red and green color blindness does not effect individual's ability to survive or procreate thus unitary fitness factors (wi=1,i=1,,5) can be assigned. Substituting the above parameters' values in system (5)-(6) the following equilibrium points are found: ˉxA=(0,0,0), ˉxB=(40231.5,37100.5,6263.7). Matrix A in system (7) is not Schur, thus model (5)-(6) will not converge to equilibrium ˉxA, that is the population will never converge to the state where all people are affected. Moreover matrix ˜A in (10) is Schur if evaluated at ˉxB. Applying Theorem (4.1), the domain of attraction Λ for equilibrium ˉxB is Λ={x|(xˉx)TH(xˉx)<1.23106}. Interestingly enough the distribution among healthy, affected and carrier people in equilibrium ˉxB is very close to the one in literature [1].


    7.2. Application 2

    Hemophilia A disease, is a hereditary bleeding disorder caused by a lack of blood clotting factor Ⅷ, a protein encoded by FⅧ gene placed on the X chromosome. It is largely an inherited disorder, that is the spontaneous gene mutation rate of the diseases (γ in our model) can be considered negligible [6]; therefore we can set γ=0. Affected males show a reduced reproduction capacity related to the severity of the disease symptoms; carrier females do not usually show any sign of the disease [9]. Due to the severity of the disease, one can assign fertility factors equal to zero to affected individuals (w2=w5=0); based on clinical observations the fertilities factors of the other classes can be assigned as follows: w1=w3=1 and w4=0.6. We studied the diffusion of the hemophilia A in the Italian population (2N in our model) that was equal to 59685228 individuals in 2012 (see [20]). Substituting the above parameters' values in system (5)-(6), the following equilibrium points are found: ˉxA=(0,0,0), ˉxB=(N,N,0). The domain of attraction of equilibrium ˉxA=(0,0,0) can be derived applying results in Proposition 1. We compute the coefficients of Lyapunov function (14) by solving the optimization problem in Section 4.2 through GA, obtaining the optimal values α=0.01, β=4.83 and μ=2.51 and we compute the domain of attraction Φ={x|x1<5.95107}. Indeed for almost any initial population distribution system (5)-(6) will converge to ˉxA; different fertility factors datas, gathered from direct population analysis, could lead to a different population distribution.


    8. Conclusions

    We developed a discrete time nonlinear model for X-linked recessive diseases aiming at describing the spread of such diseases in a population. The model accounts for de novo mutations on the inheritance pattern and distinct fitness factors. Under the assumption of constant population size we analyzed system's properties and performed stability analysis of equilibrium points through Lyapunov second method.

    Extensions of the present work should consider sensitivity analysis to parameters variations as well as the effect on the results when weakening some of the model assumptions such as the hypothesis that the population has a constant size or an equal number of females and males.

    Future developments will assign different de novo gene mutation rates to females and males as well as include de novo mutation possibly changing a defective gene into a normal one.


    Appendix

    Proof of Theorem 4.1.

    Proof. Consider the candidate Lyapunov function:

    V(y)=yTHywithy=xˉx.

    Then,

    ΔV(y)=V(y+)V(y)=[˜Ay+y1By]TH[˜Ay+y1By]yTHy=yT(˜ATH˜AH)y+2yTBTH˜AyTy1+yTBTHByy21=yTGy+2yTBTH˜Ayy1+yTBTHByy21yTGy+2y2BTH˜A|y1|+yTBTHBy|y1|2. (25)

    Since HT=H>0, we can use the following as upper bounds for terms in inequality (25):

    yTGyλ1yTHyλ1:=λmin(H1G)y2λ2yTHyλ2:=1/λmin(H)yTBTHByλ3yTHyλ3:=λmax(H1BTHB).

    Substituting these inequalities into (25) leads to:

    ΔVV(y)(λ12b|y1|λ3|y1|2) (26)

    where b:=λ2BTH˜A. One may verify that λ12b|y1|λ3|y1|2>0 if and only if |y1|<δ where

    δ:=(bλ1)2+λ1λ3bλ1>0.

    Since |y1|2=(Cy)TCy where C=[100]T and HT=H>0, exploiting properties on eigenvalues of symmetric and positive definite matrices we have

    |y1|2λ4yTHyλ4:=λmax(H1CCT).

    Hence, whenever V(y)<r=δ2/λ4 we have |y1|<δ. It now follows from (26) that

    ΔV(y)<0when0<V(y)<r.

    Conditions for matrix A to be Schur. When ˉx=(0,0,0), matrix ˜A in system (10) is equal to matrix A in (8).

    Matrix A in (8) is Schur iff :

    ˜γw24+˜γ2w15w24<2. (27)

    Proof. The characteristic polynomial of A is given by

    p(λ)=det(λIA)=λ3a13λ2a13a31λ.

    Since a13,a310, the criterion for a Schur polynomial reduces to

    a13a31<1a13<1a13a31

    which are equivalent to

    a13+a13a31xy1

    that is,

    ˜γw24+˜γ2w15w24<2.

    Conditions on wij for positiveness of system (5)-(6). In what follows we prove that system (5)-(6) is positive if the system is initialized in a positive state and if wij are chosen within an appropriate set.

    In particular we detail the proof for f1(x); analogous reasoning gives the conditions for the positiveness of f3(x) and f4(x).

    Let's assume wij1 i,j, w1w2, w3w4, γ1 and the initial states 0xi(0)N, i=1,3,4; if we rewrite (6a) as:

    f1(x)=1N[w3˜γ(w1w2)x1x3+12w4˜γ(w1w2)x1x4+w23˜γNx3+12w24˜γNx4]

    we can deduce that, holding the previous assumptions, x1(1)0. Consider now

    Nf1(x)=N˜γ2N[(2w3x3+w4x4)(Nw2+(w1w2)x1)]N˜γ2N[w3(2x3+x4)(Nw2+(w1w2)x1)]N˜γ2N[2w3N(Nw2+(w1w2)N)]=N˜γw13N (29)

    Under the previous assumptions the condition Nf1(x)0 holds, hence x1(1)N. By mathematical induction, we prove 0x1(k)N, k>0.

    Note that conditions on wij do not affect the comprehensiveness of our epidemiological model, since they simply require that affected individuals show a fitness factor less than or equal to healthy ones.


    Acknowledgments

    Authors are grateful to prof. Emanuele Durante Mangoni, MD PhD, from the Internal Medicine Unit, Second University of Naples, Italy, for helpful discussions and literature suggestions on X-linked diseases and their inheritance pattern and for his critical review of the manuscript.


    [1] [ M. AG and D. SS., The Metabolic and Molecular Bases of Inherited Disease, chapter Color vision and its genetic defects, 5955–76, McGraw-Hill, 2001.
    [2] [ L. Allen,D. Thrasher, The effects of vaccination in an age-dependent model for varicella and herpes zoster, IEEE Transactions on Automatic Control, 43 (1998): 779-789.
    [3] [ A. Aswani and C. Tomlin, Computer-aided drug discovery for pathway and genetic diseases, in Proc. 49th IEEE Conference on Decision and Control 2010, 2010,4709–4714
    [4] [ E. August, G. Craciun and H. Koeppl, Finding invariant sets for biological systems using monomial domination, in CDC, 2012,3001–3006
    [5] [ C. F. Baer, Does mutation rate depend on itself PLoS Biol, 6 (2008), e52.
    [6] [ J. Becker,R. Schwaab,A. Möller-Taube,U. Schwaab,W. Schmidt,H. H. Brackmann,T. Grimm,K. Olek,J. Oldenburg, Characterization of the factor viii defect in 147 patients with sporadic hemophilia a: Family studies indicate a mutation type-dependent sex ratio of mutation frequencies, American Journal of Human Genetics, 58 (1996): 657-670.
    [7] [ D. Bick, Engineering in genomics-genetic disease diagnosis: Challenges and opportunities, IEEE Engineering in Medicine and Biology Magazine, 14 (1995): 226-228.
    [8] [ F. Blanchini,E. Franco,G. Giordano, Determining the structural properties of a class of biological models, CDC, 2 (2012): 5505-5510.
    [9] [ D. J. Bowen, Haemophilia a and haemophilia b: Molecular insights, Molecular Pathology, 55 (2002): 1-18.
    [10] [ C. Cannings, Equilibrium, convergence and stability at a sex-linked locus under natural selection, Genetics, 56 (1967): 613-618.
    [11] [ J. Chunyan,J. Daqing,Y. Q. Yang,S. Ningzhong, Dynamics of a multigroup SIR epidemic model with stochastic perturbation, Automatica, 48 (2012): 121-131.
    [12] [ J. F. Crow, How much do we know about spontaneous human mutation rates?, Environmental and Molecular Mutagenesis, 21 (1993), 122–129, URL http://dx.doi.org/10.1002/em.2850210205.
    [13] [ J. F. Crow, The origins, patterns and implications of human spontaneous mutation, Nature Reviews Genetics, 1 (2000): 40-47.
    [14] [ C. Del Vecchio, L. Glielmo and M. Corless, Equilibrium and stability analysis of x-chromosome linked recessive diseases model, in Proc. IEEE 51st Annual Conference on Decision and Control (CDC), 2012,4936–4941
    [15] [ C. Del Vecchio, L. Glielmo and M. Corless, Non linear discrete time epidemiological model for x-linked recessive diseases in 22nd Mediterranean Conference on Control and Automation, June 2014, Palermo, Italy 2014.
    [16] [ A. W. F. Edwards, Foundations of Mathematical Genetics Cambridge University Press, 2000.
    [17] [ W. J. Ewens, Mathematical Population Genetics Ⅰ: Theoretical Introduction, Springer, 2004.
    [18] [ D. P. Germain, General aspects of x-linked diseases, in Fabry Disease: Perspectives from 5 Years of FOS, Oxford PharmaGenesis, 2006.
    [19] [ H. Hethcote, The mathematics of infectious diseases, SIAM Review, 42 (2000): 599-653.
    [20] [ http://www.istat.it/it/charts/popolazioneresidente, Authors' last visit on February 2015.
    [21] [ http://www.who.int/genomics/public/geneticdiseases/en/index2.html, Authors' last visit on February 2015.
    [22] [ C. Hyeygjeon,A. Astolfi, Control of HIV infection dynamics, IEEE Control Systems Magazine, 28 (2008): 28-39.
    [23] [ B. J., Worldwide prevalence of red-green color deficiency., Journal of the Optical Society of America, 29 (2012): 313-320.
    [24] [ M. J. Keeling and P. Rohani, Modeling Infectious Diseases in Humans and Animals Princeton University Press, 2008.
    [25] [ M. J. Khoury, T. H. Beaty and B. H. Cohen, Fundamentals of Genetic Epidemiology, Oxford University Press, 1993.
    [26] [ M. Lachowicz and J. Miekisz, From Genetics to Mathematics vol. 79 of Series on Advances in Mathematics for Applied Sciences, World Scientific Publications, 2009.
    [27] [ K. Lange, Mathematical and Statistical Methods for Genetic Analysis, Second Edition, Springer, New York, 2002.
    [28] [ K. Lange, Calculation of the equilibrium distribution for a deleterious gene by the finite fourier transform, Biometrics, 38 (1982): 79-86.
    [29] [ K. Lange,B. Redelings, Disease gene dynamics in a population isolate, An Introduction to Mathematical Modeling in Physiology, Cell Biology, and Immunology: American Mathematical Society, Short Course, January 8-9,2001, New Orleans, Louisiana, 59 (2002): 119-138.
    [30] [ D. Luenberger, Introduction to Dynamic Systems, John Wiley and Sons, 1979.
    [31] [ T. Nagylaki, Selection and mutation at an x-linked locus, Annals of Human Genetics, 41 (1977): 241-248.
    [32] [ T. Nagylaki, Introduction to Theoretical Population Genetics, Biomathematics, 21. Springer-Verlag, Berlin, 1992.
    [33] [ H. A. Orr, Fitness and its role in evolutionary genetics, Nature reviews. Genetics, 10 (2009): 531-539.
    [34] [ R. H. Post, Population differences in red and green color vision deficiency: A review, and a query on selection relaxation, Social Biology, 29 (1982): 299-315.
    [35] [ D. B. Saakian,Z. Kirakosyan,C.-K. Hu, Diploid biological evolution models with general smooth fitness landscapes and recombination, Physical Review E, 77 (2008): 061907.
    [36] [ G. D. Smith,S. Ebrahim,S. Lewis,A. L. Hansell,L. J. Palmer,P. R. Burton, Genetic epidemiology and public health: Hope, hype, and future prospects, The Lancet, 366 (2005): 1484-1498.
    [37] [ A. E. Stark, Determining the frequency of sporadic cases of rare X-linked disorders, Annals of Translational Medicine, 4 (2016).
    [38] [ B. Sturmfels, Polynomial equations and convex polytopes, The American Mathematical Monthly, 105 (1998): 907-922.
    [39] [ J. M. Szucs, Selection and mutation at a diallelic X-linked locus, Journal of Mathematical Biology, 29 (1991): 587-627.
    [40] [ R. M. Winter, Estimation of male to female ratio of mutation rates from carrier-detection tests in x-linked disorders, American Journal of Human Genetics, 32 (1980): 582-588.
    [41] [ X. Xia, Estimation of HIV/AIDS parameters, Automatica, 39 (2003): 1983-1988.
    [42] [ N. Yasuda,K. Kondô, No sex difference in mutations rates of duchenne muscular dystrophy, Journal of Medical Genetics, 17 (1980): 106-111.
    [43] [ L. Yeghiazarian,M. Kaiser, Markov model of sex-linked recessive trait transmission, Mathematical and Computer Modelling, 29 (1999): 71-81.
    [44] [ M. Yousefi,A. Datta,E. Dougherty, Optimal intervention in markovian gene regulatory networks with random-length therapeutic response to antitumor drug, Biomedical Engineering, IEEE Transactions on, 60 (2013): 3542-3552.
  • This article has been cited by:

    1. Julie Jensen DelFavero, Amy J. Jnah, Desi Newberry, Glucose-6-Phosphate Dehydrogenase Deficiency and the Benefits of Early Screening, 2020, 39, 0730-0832, 270, 10.1891/0730-0832.39.5.270
    2. Carmen Del Vecchio, Francesca Verrilli, Luigi Glielmo, Modelling and stability analysis in human population genetics with selection and mutation, 2018, 41, 01704214, 1492, 10.1002/mma.4679
    3. C. Del Vecchio, F. Verrilli, L. Glielmo, When sex matters: a complete model of X-linked diseases, 2018, 47, 0308-1079, 549, 10.1080/03081079.2018.1473391
  • Reader Comments
  • © 2017 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(4401) PDF downloads(617) Cited by(3)

Article outline

Figures and Tables

Figures(7)  /  Tables(4)

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog