Processing math: 77%
Research article Special Issues

Predicting S-nitrosylation proteins and sites by fusing multiple features


  • Protein S-nitrosylation is one of the most important post-translational modifications, a well-grounded understanding of S-nitrosylation is very significant since it plays a key role in a variety of biological processes. For an uncharacterized protein sequence, it is a very meaningful problem for both basic research and drug development when we can firstly identify whether it is a S-nitrosylation protein or not, and then predict the specific S-nitrosylation site(s). This work has proposed two models for identifying S-nitrosylation protein and its PTM sites. Firstly, three kinds of features are extracted from protein sequence: KNN scoring of functional domain annotation, PseAAC and bag-of-words based on the physical and chemical properties of amino acids. Secondly, the synthetic minority oversampling technique is used to balance the data sets, and some state-of-the-art classifiers and feature fusion strategies are performed on the balanced data sets. In the five-fold cross-validation for predicting S-nitrosylation proteins, the results of Accuracy (ACC), Matthew's correlation coefficient (MCC) and area under ROC curve (AUC) are 81.84%, 0.5178, 0.8635, respectively. Finally, a model for predicting S-nitrosylation sites has been constructed on the basis of tripeptide composition (TPC) and the composition of k-spaced amino acid pairs (CKSAAP). To eliminate redundant information and improve work efficiency, elastic nets are employed for feature selection. The five-fold cross-validation tests have indicated the promising success rates of the proposed model. For the convenience of related researchers, the web-server named "RF-SNOPS" has been established at http://www.jci-bioinfo.cn/RF-SNOPS

    Citation: Wang-Ren Qiu, Qian-Kun Wang, Meng-Yue Guan, Jian-Hua Jia, Xuan Xiao. Predicting S-nitrosylation proteins and sites by fusing multiple features[J]. Mathematical Biosciences and Engineering, 2021, 18(6): 9132-9147. doi: 10.3934/mbe.2021450

    Related Papers:

    [1] Jianxin Yang, Zhipeng Qiu, Xue-Zhi Li . Global stability of an age-structured cholera model. Mathematical Biosciences and Engineering, 2014, 11(3): 641-665. doi: 10.3934/mbe.2014.11.641
    [2] Mingtao Li, Xin Pei, Juan Zhang, Li Li . Asymptotic analysis of endemic equilibrium to a brucellosis model. Mathematical Biosciences and Engineering, 2019, 16(5): 5836-5850. doi: 10.3934/mbe.2019291
    [3] Linhua Zhou, Meng Fan, Qiang Hou, Zhen Jin, Xiangdong Sun . Transmission dynamics and optimal control of brucellosis in Inner Mongolia of China. Mathematical Biosciences and Engineering, 2018, 15(2): 543-567. doi: 10.3934/mbe.2018025
    [4] Zongmin Yue, Yuanhua Mu, Kekui Yu . Dynamic analysis of sheep Brucellosis model with environmental infection pathways. Mathematical Biosciences and Engineering, 2023, 20(7): 11688-11712. doi: 10.3934/mbe.2023520
    [5] Mingtao Li, Guiquan Sun, Juan Zhang, Zhen Jin, Xiangdong Sun, Youming Wang, Baoxu Huang, Yaohui Zheng . Transmission dynamics and control for a brucellosis model in Hinggan League of Inner Mongolia, China. Mathematical Biosciences and Engineering, 2014, 11(5): 1115-1137. doi: 10.3934/mbe.2014.11.1115
    [6] Zhisheng Shuai, P. van den Driessche . Impact of heterogeneity on the dynamics of an SEIR epidemic model. Mathematical Biosciences and Engineering, 2012, 9(2): 393-411. doi: 10.3934/mbe.2012.9.393
    [7] Yaoyao Qin, Xin Pei, Mingtao Li, Yuzhen Chai . Transmission dynamics of brucellosis with patch model: Shanxi and Hebei Provinces as cases. Mathematical Biosciences and Engineering, 2022, 19(6): 6396-6414. doi: 10.3934/mbe.2022300
    [8] Qiuyi Su, Jianhong Wu . Impact of variability of reproductive ageing and rate on childhood infectious disease prevention and control: insights from stage-structured population models. Mathematical Biosciences and Engineering, 2020, 17(6): 7671-7691. doi: 10.3934/mbe.2020390
    [9] Pengyan Liu, Hong-Xu Li . Global behavior of a multi-group SEIR epidemic model with age structure and spatial diffusion. Mathematical Biosciences and Engineering, 2020, 17(6): 7248-7273. doi: 10.3934/mbe.2020372
    [10] Tsanou Berge, Samuel Bowong, Jean Lubuma, Martin Luther Mann Manyombe . Modeling Ebola Virus Disease transmissions with reservoir in a complex virus life ecology. Mathematical Biosciences and Engineering, 2018, 15(1): 21-56. doi: 10.3934/mbe.2018002
  • Protein S-nitrosylation is one of the most important post-translational modifications, a well-grounded understanding of S-nitrosylation is very significant since it plays a key role in a variety of biological processes. For an uncharacterized protein sequence, it is a very meaningful problem for both basic research and drug development when we can firstly identify whether it is a S-nitrosylation protein or not, and then predict the specific S-nitrosylation site(s). This work has proposed two models for identifying S-nitrosylation protein and its PTM sites. Firstly, three kinds of features are extracted from protein sequence: KNN scoring of functional domain annotation, PseAAC and bag-of-words based on the physical and chemical properties of amino acids. Secondly, the synthetic minority oversampling technique is used to balance the data sets, and some state-of-the-art classifiers and feature fusion strategies are performed on the balanced data sets. In the five-fold cross-validation for predicting S-nitrosylation proteins, the results of Accuracy (ACC), Matthew's correlation coefficient (MCC) and area under ROC curve (AUC) are 81.84%, 0.5178, 0.8635, respectively. Finally, a model for predicting S-nitrosylation sites has been constructed on the basis of tripeptide composition (TPC) and the composition of k-spaced amino acid pairs (CKSAAP). To eliminate redundant information and improve work efficiency, elastic nets are employed for feature selection. The five-fold cross-validation tests have indicated the promising success rates of the proposed model. For the convenience of related researchers, the web-server named "RF-SNOPS" has been established at http://www.jci-bioinfo.cn/RF-SNOPS



    Brucellosis, a zoonotic disease, is a natural epidemic disease that is not only prevalent among livestock and humans but also widely spread in wild animals [1,2]. There are human and animal brucellosis in most countries of the world, and high-risk areas of epidemics are mainly distributed in developing countries, such as Syria, Jordan, Zambia, Mongolia [3,4,5,6]. It has brought huge losses to the livestock industry worldwide and serious health problems to livestock-related practitioners, there are hundreds of thousands of new cases reported annually [6,7,8]. Infected domestic and wild animals and their excreta are the main source of infection, it is contact and pathogen infection which are the main modes of brucellosis transmission [9]. Human brucellosis is rarely transmitted to susceptible animals, and there is no infection between people reported [10,11]. Therefore, it is the eradication of animal brucellosis that is the only way to solving human health problem, and understanding the mechanisms and risk factors of the spread of brucellosis is one of the first problems that must be solved.

    Brucellosis is transmitted to susceptible individuals mainly by contact with infectious individuals or by sucking pathogens from the environment [8]. In animals, the transmission of brucellosis mainly occurs between sexually mature animals, young animals may also be infected, but generally do not have any clinical manifestations and serological tests are usually non-positive, and the infection between young animals is so little as to be almost invisible [12]. In other words, sexually mature animals are very susceptible to brucellosis, latent infections may be found in young animals, but they are generally resistant [9]. It is important that many infected animals have a longer incubation period and that these animals remain serologically non-positive during this period. That is to say, latent animals may not be infectious and it is almost impossible to be found through detection, which is an important risk factor in the elimination of brucellosis [9]. Therefore, the impact of these mechanisms on brucellosis is worthy to be studied using mathematical models.

    Statistical methods such as descriptive statistics, correlation analysis and time series analysis have been widely used for quantitative assessment of risk management measures for brucellosis, among which is very worthy of concern about the study of bison, elk and livestock brucellosis (see [13,14,15,16,17]). Theoretical studies on the impacts of the transmission mechanisms of brucellosis and applied studies on the assessments of risk management measures have also been studied using the kinetic model (see [18,19,20,21,22]). Especially in recent years, many kinetic models with indirect transmission have been established to analyze brucellosis transmission (see [23,24,25,26,27]). Although there have been many studies on brucellosis, there are still many transmission mechanisms and risk factors that are not considered in existing models. For example, infected individuals may have no infectivity in the early stage, during which time is different; and animals that are not sexually mature are hardly infected with direct and indirect modes [9]. Therefore, in the present work, a multi-stage dynamic model with distributed time delay is proposed involving the above risk factors and general nonlinear incidences. The existence and uniqueness of the endemic equilibrium is analyzed, and the local and global asymptotic stability of equilibria is proved.

    The rest of this manuscript is constructed as follows. The dynamic model with distributed time delay and some preliminary results are given in Section 2. In Section 3, the content of research is the global dynamics of the disease-free equilibrium. The global stability of the equilibrium of persistent infection are analyzed in Section 4. In Section 5, the stability results are further explained by numerical simulation. A summary and further discussion is proposed in Section 6.

    According to the transmission mechanisms of animal brucellosis, we classify animal population into three compartments: The young susceptible compartment S1(t), the adult susceptible compartment S2(t) (sexually mature), the infected compartment I(t). Similar to the definition in literature [25], B(t) can be defined as the concentration or number of pathogens in the environment. Some explanations and assumptions about the kinetic model are listed as follows. (Ⅰ) Since brucellosis can cause abortion and significantly reduce the survival rate of young animals, we suppose that infected animals have no birth rate. (Ⅱ) The supplementary rate of young susceptible animal population S1(t) is mainly derived from birth and import, it is assumed that the supplementary rate is A+bS2(t). (Ⅲ) Usually, there are two different stages for infected animals, which are no infection force in the early stage and are infectious carriers when infected animals begin to shed brucella. Therefore, we make use of a delay τ to describe the time from a susceptible individual to a infectious individual, and since the length of this time is varied, thus the delay τ is a distributed parameter in the interval [0,h], where h is the maximum value of the delay. As a result, the susceptible class S2(t) is reduced due to infection at the rate S2g(B)+S2h0ρ(τ)f(I(tτ))e(μ2+c)τdτ. (Ⅳ) Since the infected animal sheds the pathogen after the time τ, then the pathogen shedding rate of the infected animal is h0ρ(τ)h(I(tτ))e(μ2+c)τdτ. (Ⅴ) In animal breeding, only the basic ewes breed newborns, and adult animals are widely used for slaughter or trade, then the birth rate b of adult animals is assumed to be less than the elimination rate μ2 of adult animals. Therefore, the modelling for animal brucellosis is given through the following distributed time-delay system:

    {dS1dt=A+bS2μ1S1σ1S1,dS2dt=σ1S1S2g(B)S2h0ρ(τ)f(I(tτ))e(μ2+c)τdτμ2S2,dIdt=S2h0ρ(τ)f(I(tτ))e(μ2+c)τdτ+S2g(B)(μ2+c)I,dBdt=h0ρ(τ)h(I(tτ))e(μ2+c)τdτdB. (2.1)

    Here, b is the birth rate of adult animals. μ1 and μ2 are the elimination rates of young and adult animals, respectively. σ1 is the transfer rate from young susceptible individuals to sexually mature individuals. The culling rate is defined by c and d is the decay rate of the pathogen. f(I) and g(B) are contact and indirect infection rates, respectively. ρ(τ) is a distributed function, it is non-negative, continuous and h0ρ(τ)dτ=1.

    The initial conditions for system (2.1) are given as follows:

    {S1(x)=ϕ1(x),S2(x)=ϕ2(x),I(x)=ϕ3(x),B(x)=ϕ4(x),x[h,0],h>0,ϕ=(ϕ1,ϕ2,ϕ3,ϕ4)C+C. (2.2)

    Here, C denotes the Banach space C([h,0],4) of continuous functions mapping the interval [h,0] into 4 with the sup-norm ϕ=supx[h,0]|ϕi(x)|,i=1,2,3,4 for ϕC. The nonnegative cone of C is defined as C+=C([h,0],4+).

    In order to make epidemiological significance for system (2.1), f,g and h are assumed to be second-order continuous differentiable functions and satisfy the following hypotheses:

    (H1) f(0)=g(0)=h(0)=0 and f(I),g(B),h(I)>0 for I,B>0;

    (H2) f(I),g(B)>0 and f(I),g(B)0 for B,I0;

    (H3) h(I)>0 and h(I)0 for I0.

    The function f may be βIp or saturation incidences kln(1+λIk) and βIp1+kIp with constants β,p,λ,k>0 [28,29]. The pathogen infection rate g may be λB1+TB or η(1eαB) with constants λ,η,α>0 and T0 [30]. The function h may be kI with constants k>0 [31].

    It is easy to verify that the system (2.1) always has a disease-free equilibrium E0=(S01,S02,0,0), where

    S01=Aμ2μ2(μ1+σ1)bσ1,S02=Aσ1μ2(μ1+σ1)bσ1.

    The reproduction number of sytem (2.1) is given by the following expression:

    R0=nS02fI(0)μ2+c+nS02gB(0)hI(0)d(μ2+c)=R01+R02. (2.3)

    where n=h0ρ(τ)e(μ2+c)τdτ. e(μ2+c)τ denotes the probability of survival from a newly infected susceptible individual to an infectious individual. n is the total survival rate of infected individuals. fI(0) is the infection rate by an infectious individual. gB(0) is the infection rate of brucella. 1μ2+c is the average life span of infected individual. hI(0)d represents the total number of brucella shed by infectious individuals. Therefore, based on the explanation in the literature [26], R0 relies on direct and indirect infection and can be divided into two parts R01 and R02.

    In the next section, we establish that the solution of system (2.1) is non-negative and bounded and analyze the uniqueness of the equilibrium point of persistent infection. The following qualitative results indicate the nonnegativity and boundedness of the solution of system (2.1) with initial conditions (2.2).

    Theorem 2.1. (S1(t),S2(t),I(t),B(t)) is the solution of system (2.1) with the initial conditions (2.2), then S1(t),S2(t),I(t),B(t) are nonnegative and ultimately bounded.

    Proof. Since (˙S1(t)+˙S2(t))=A>0 for t[0,+) when S1(t)=S2(t)=0, it implies that S1(t)+S2(t)0 for all t[0,+). Therefore, if S1(t)=0, then S2(t)0 and ˙S1(t)=A+bS2>0, so that S1(t)0. We denote

    a1(t)=h0ρ(τ)f(I(tτ))e(μ2+c)τdτ+g(B)+μ2,a2(t)=S2(t)(h0ρ(τ)f(I(tτ))e(μ2+c)τdτ+g(B)),

    and

    a3(t)=h0ρ(τ)h(I(tτ))e(μ2+c)τdτ.

    From the last three equations in (2.1), it can conclude that

    S2(t)=S2(0)et0a1(η)dη+σ1t0etηa1(ξ)dξS1(η)dη0,I(t)=I(0)e(μ2+c)t+t0e(μ2+c)(tη)a2(η)dη0,

    and

    B(t)=B(0)edt+t0ed(tη)a3(η)dη0

    for all t0. Thus, S1(t),S2(t),I(t),B(t)0 for t0.

    We now analyze the boundedness of the solution of system (2.1). In fact,

    (˙S1(t)+˙S2(t)+˙I(t))Aμ1S1(μ2b)S2(μ2+c)I
    Aμ(S1(t)+S2(t)+I(t)),

    where μ=min{μ1,μ2b}.

    Hence, lim supt(S1(t)+S2(t)+I(t))Aμ, it implies that lim suptI(t)Aμ and lim suptB(t)ndh(Aμ). Therefore, S1(t),S2(t),I(t),B(t) are ultimately bounded.

    Since the solution of system (2.1) is nonnegative and ultimately bounded, the set

    Ω={(S1(),S2(),I(),B())C+:S1+S2+IAμ,Bndh(Aμ)},

    is positively invariant for system (2.1).

    In order to analyze the uniqueness of the positive solution of system (2.1), the following lemma is given:

    Lemma 2.1. Assume that conditions (H1)(H3) are satisfied, the functions f(I)I, h(I)I, g(B)B and g(h(I))I are monotonic decreasing for I,B>0.

    Proof. Since f(I)0, it shows that f(I) is monotonic decreasing, it follows that

    f(I)I=f(I)f(0)I0=f(ξ1)f(I),ξ1(0,I),

    and

    (f(I)I)=f(I)If(I)I20.

    That is to say, f(I)I is a monotonically decreasing function. On the basis of the above method, it can show that h(I)I and g(B)B are also monotonic decreasing. Noting that

    g(h(I))I=g(h(I))g(h(0))h(I)h(0)h(I)I=g(h(ξ2))h(ξ3)g(h(I))h(I),ξ2,ξ3(0,I),

    It can deduce that

    (g(h(I))I)=g(h(I))h(I)Ig(h(I))I20.

    Therefore, g(h(I))I is also monotonic decreasing.

    For system (2.1), the endemic equilibrium E=(S1,S2,I,B) can be derived from the following algebraic equations:

    {A+bS2=μ1S1+σ1S1,σ1S1=S2(nf(I)+g(B))+μ2S2,S2(nf(I)+g(B))=(μ2+c)I,nh(I)=dB. (2.4)

    By direct calculation, it can be written as

    {S1=A+bS2μ1+σ1,m(S2)=S2(nf(I)+g(B)),S2(nf(I)+g(B))=(μ2+c)I,B=H(I).

    where m(S2)=Aσ1(μ2(μ1+σ1)bσ1)S2μ1+σ1 and H(I)=ndh(I).

    Let us define

    F1(S2,I)m(S2)S2(nf(I)+g(H(I))),F2(S2,I)S2(nf(I)+g(H(I)))(μ2+c)I.

    Similar to the method of analysis in literature [26], using Lemma 2.1, the following result can be summarized:

    Theorem 2.2. Assume that conditions (H1)(H3) hold. Then there is a unique positive solution E=(S1,S2,I,B) of system (2.1) if and only if R0>1.

    In this section, we show that the global stability of the equilibrium point E0 of system (2.1) is independent of the initial value. The following conclusions are first obtained.

    Lemma 3.1. Assume that conditions (H1)(H3) hold. The disease-free equilibrium E0 is locally asymptotically stable if R01 and is unstable if R0>1.

    Proof. The characteristic equation at E0 is

    |λ+μ1+σ1b00σ1λ+μ2S02fI(0)Γ(λ)S02gB(0)00λ+μ2+cΓ(λ)S02fI(0)S02gB(0)00hI(0)Γ(λ)λ+d|=0, (3.1)

    where

    Γ(λ)=h0ρ(τ)e(μ2+c+λ)τdτ.

    It follows from (3.1) that

    (λ+μ1+σ1)(λ+μ2)((λ+d)(λ+μ2+cΓ(λ)S02fI(0))Γ(λ)S02gB(0)hI(0))=bσ1((λ+d)(λ+μ2+cΓ(λ)S02fI(0))Γ(λ)S02gB(0)hI(0)).

    That is,

    (λ+μ1+σ1)(λ+μ2)(λ+d)(λ+μ2+c)bσ1(λ+d)(λ+μ2+c)
    =((λ+μ1+σ1)(λ+μ2)bσ1)((λ+d)Γ(λ)S02fI(0)+Γ(λ)S02gB(0)hI(0)).

    It follows that

    H1(λ)((λ+d)(λ+μ2+c)(λ+d)Γ(λ)S02fI(0)Γ(λ)S02gB(0)hI(0))=0,

    where H1(λ)=(λ+μ1+σ1)(λ+μ2)bσ1. Since μ1+σ1+μ2>0 and μ2(μ1+σ1)bσ1>0, then H1(λ) consists of two roots which are negative real parts. Therefore, we only analyze the distribution of the roots of the following equation:

    H2(λ)=(λ+d)(λ+μ2+c)(λ+d)Γ(λ)S02fI(0)Γ(λ)S02gB(0)hI(0)=0. (3.2)

    Assume now that R0>1, then

    H2(0)=d(μ2+c)dnS02fI(0)nS02gB(0)hI(0))<0,H2(+)=+.

    Hence H2(λ) has at least one positive root in [0,+), then E0 is unstable if R0>1.

    From (3.2), we have

    (λ+d)(λ+μ2+c)=Γ(λ)(λS02fI(0)+dS02fI(0)+S02gB(0)hI(0))=(μ2+c)F(λ)n(λR01+dR0),

    or

    (λ+d)(λμ2+c+1)=R0F(λ)n(λR01R0+d). (3.3)

    Next, considering the case R01. If λ=x+yi is a solution of (3.3), one shows that x<0. Otherwise, x0 implies

    |λ+d|>|λR01R0+d|,|λμ2+c+1|>1,|R0F(λ)n|1,

    and thus

    |(λ+d)(λμ2+c+1)|>|R0F(λ)n(λR01R0+d)|,

    this is a contradiction to (3.3). Therefore, all roots of Eq (3.3) have no zero and positive real parts when R01, this shows that E0 is locally asymptotically stable.

    Theorem 3.1. Assume that conditions (H1)(H3) are established. If R01, the disease-free equilibrium E0=(S01,S02,0,0) of system (2.1) is globally asymptotically stable.

    Proof. Since the functions f(I)I, g(B)B and h(I)I are decreasing, then we have

    nS2f(I)(μ2+c)IlimI0+nS02f(I)(μ2+c)I=nS02fI(0)μ2+cb1,S2g(B)dBlimB0+S02g(B)dB=S02gB(0)db2,nh(I)(μ2+c)IlimI0+nh(I)(μ2+c)I=nhI(0)μ2+c.

    Define

    J=(10nhI(0)μ2+c1),(a1,a2)=(b1,b2)J1.

    We find a1=R0 and define a Lyapunov functional L as follows:

    L(t)=L1(t)+L2(t)+L3(t),

    where

    L1=R0(S1S01S01lnS1S01+S2S02S02lnS2S02+I)+a2B,L2=R0S02h0Ψ(τ)f(I(tτ))dτ,L3=a2h0Ψ(τ)h(I(tτ))dτ,

    and

    Ψ(τ)=hτφ(s)ds,φ(s)=ρ(s)e(μ2+c)s.

    Then the derivative of L1 along the positive solutions of system (2.1) is

    dL1dt=R0(1S01S1)dS1dt+R0(1S02S2)dS2dt+R0dIdt+a2dBdt=R0(2A+bS02+σ1S01μ1S1(μ2b)S2S01AS1bS01S2S1σ1S02S1S2)+R0(S02g(B)+S02h0φ(τ)f(I(tτ))dτ(μ2+c)I)+a2(h0φ(τ)h(I(tτ))dτdB). (3.4)

    Calculating the derivative of L2(t) along the solutions of system (2.1), one obtains

    dL2dt=R0S02h0Ψ(τ)df(I(tτ))dtdτ=R0S02h0Ψ(τ)df(I(tτ))dτdτ=R0S02(Ψ(τ)f(I(tτ)))|h0+R0S02h0dΨ(τ)dτf(I(tτ))dτ=R0S02nf(I(t))R0S02h0φ(τ)f(I(tτ))dτ. (3.5)

    Similar to the above-used method, it can obtain that

    dL3dt=a2nh(I(t))a2h0φ(τ)h(I(tτ))dτ. (3.6)

    Combining the Eqs (3.4), (3.5) and (3.6), it follows that

    dLdt=dL1dt+dL2dt+dL3dt=R0μ1S01(2S1S01S01S1)+R0bS02(2S1S02S01S2S01S2S1S02)+R0(μ2b)S02(3S2S02S1S02S01S2S01S1)+R0(nS02f(I)(μ2+c)I,S02g(B)dB)((μ2+c)I,dB)T(R0,a2)(10nh(I)(μ2+c)I1)((μ2+c)I,dB)TR0(nS02fI(0)(μ2+c),S02gB(0)d)((μ2+c)I,dB)T(R0,a2)(10nhI(0)μ2+c1)((μ2+c)I,dB)T=(R01)(b1,b2)((μ2+c)I,dB)T0.

    Similar to the analysis of Theorem 1 in the literature [32], it follows from Lemma 3.1 that the disease-free equilibrium E0 is globally asymptotically stable by LaSalle's Invariance Principle [33].

    According to Lemma 3.1, the disease-free steady state solution E0 is unstable when R0>1. Using Theorem 4.2 in [34], the uniform permanence of system (2.1) can be proven, the process is ignored here. In this following section, by constructing a Lyapunov functional, the global stability of the equilibrium of persistent infection is proved. We first analyze its local stability.

    Lemma 4.1. Assume that conditions (H1)(H3) are ture, If R0>1, the endemic equilibrium E of system (2.1) exists and is locally asymptotically stable.

    Proof. For system (2.1), the characteristic equation at E is

    |λ+μ1+σ1b00σ1H3(λ)Γ(λ)S2fI(I)S2gB(B)0nf(I)g(B)H4(λ)S2gB(B)00Γ(λ)hI(I)λ+d|=0,

    where H3(λ)=λ+nf(I)+g(B)+μ2, H4(λ)=λ+μ2+cΓ(λ)S2fI(I). By simple calculation, one obtains

    (λ+d)(λ+μ2+c)Φ1(λ)=Φ2(λ)Γ(λ)((λ+d)S2fI(I)+hI(I)S2gB(B))=Φ2(λ)Γ(λ)nΛn(λS2fI(I)Λ+d),

    or

    (λ+d)(λμ2+c+1)Φ1(λ)=Φ2(λ)Γ(λ)nΛnμ2+c(λS2fI(I)Λ+d), (4.1)

    where

    Φ1(λ)=(λ+μ1+σ1)(λ+μ2+nf(I)+g(B))bσ1,Φ2(λ)=(λ+μ1+σ1)(λ+μ2)bσ1,Λ=S2fI(I)+1dS2hI(I)gB(B)

    Noting that from Lemma 1

    Λnμ2+c=S2fI(I)n+1dS2hI(I)gB(B)nμ2+cS2(f(I)nI+ndh(I)Ig(B)B)μ2+c=S2(f(I)n+g(B))I(μ2+c)=1.

    Assume λ=x+yi is a solution of (4.1). If x0, we then have

    |λ+d|>|λS2fI(I)Λ+d|,|λμ2+c+1|>1,|F(λ)n|1,

    and

    |Φ1(λ)Φ2(λ)|=|(λ+μ1+σ1)(λ+μ2+nf(I)+g(B))bσ1(λ+μ1+σ1)(λ+μ2)bσ1|=|λ+μ2+nf(I)+g(B)bσ1λ+μ1+σ1λ+μ2bσ1λ+μ1+σ1|=|M1+nf(I)+g(B)+M2iM1+M2i|>1,

    where M1=x+μ2bσ1(x+μ1+σ1)(x+μ1+σ1)2+y2>0, M2=y+bσ1y(x+μ1+σ1)2+y2. So it concludes that

    |(λ+d)(λμ2+c+1)Φ1(λ)|>|Φ2(λ)F(λ)nΛnμ2+c(λS2fI(I)Λ+d)|,

    which is a contradiction to (4.1). Therefore, the Eq (4.1) can not have any roots with a nonnegative real part, this implies that E is locally asymptotically stable if R0>1.

    Theorem 4.1. Assume that conditions (H1)(H3) hold. If R0>1, the endemic equilibrium E=(S1,S2,I,B) of system (2.1) is globally asymptotically stable.

    Proof. Define

    L1=S1S1S1lnS1S1+S2S2S2lnS2S2+IIIlnII+S2g(B)nh(I)(BBBlnBB).

    Finding the time derivative of L1 along the positive solutions of system (2.1) gives

    dL1dt=(1S1S1)dS1dt+(1S2S2)dS2dt+(1II)dIdt+S2g(B)nh(I)(1BB)dBdt=2A+bS2+σ1S1μ1S1(μ2b)S2S1AS1bS1S2S1σ1S2S1S2+S2g(B)+S2h0φ(τ)f(I(tτ))dτ(μ2+c)IIIS2g(B)IIS2h0φ(τ)f(I(tτ))dτ+S2g(B)nh(I)(h0φ(τ)h(I(tτ))dτdB+dB)S2g(B)nh(I)BBh0φ(τ)h(I(tτ))dτ.

    By the Eq (2.4), we get

    dL1dt=μ1S1(2S1S1S1S1)+(μ2b)S2(3S2S2S1S2S1S2S1S1)+bS2(2S1S2S1S2S1S2S1S2)+nS2f(I)(3S1S1S1S2S2S1II)+S2g(B)(3+g(B)g(B)S1S1S1S2S2S1IIS2g(B)IS2g(B)I)+S2h0φ(τ)f(I(tτ))dτIIS2h0φ(τ)f(I(tτ))dτ+S2g(B)(h(I)h(I)BBBh(I)Bh(I)+1)+S2g(B)nh(I)(h0φ(τ)h(I(tτ))dτnh(I)+BBnh(I))S2g(B)nh(I)BBh0φ(τ)h(I(tτ))dτ. (4.2)

    Since the function ν(x)=1x+lnx is nonpositive for x>0 and ν(x)=0 if and only if x=1, so we define

    L2=S2f(I)h0Ψ(τ)ν(f(I(tτ))f(I))dτ.

    A direct calculation shows that

    dL2dt=S2f(I)h0Ψ(τ)ddtν(f(I(tτ))f(I))dτ=S2f(I)h0Ψ(τ)ddτν(f(I(tτ))f(I))dτ=S2f(I)(Ψ(τ)ν(f(I(tτ))f(I))h0+h0φ(τ)ν(f(I(tτ))f(I))dτ)=S2f(I)h0φ(τ)ν(f(I(t))f(I))dτ+S2f(I)h0φ(τ)ν(f(I(tτ))f(I))dτ=S2f(I)h0φ(τ)(f(I)f(I)f(I(tτ))f(I)+lnf(I(tτ))f(I))dτ=nS2f(I)f(I)f(I)+S2f(I)h0φ(τ)(f(I(tτ))f(I)+lnf(I(tτ))f(I))dτ. (4.3)

    Define

    L3=S2g(B)nh0Ψ(τ)ν(h(I(tτ))h(I))dτ.

    Calculating the time derivative of L3(t), one obtains

    dL3dt=S2g(B)nh0Ψ(τ)ddtν(h(I(tτ))h(I))dτ=S2g(B)nh0Ψ(τ)ddτν(h(I(tτ))h(I))dτ=S2g(B)nh0φ(τ)(h(I)h(I)h(I(tτ))h(I)+lnh(I(tτ))h(I))dτ=S2g(B)h(I)h(I)+S2g(B)nh0φ(τ)(h(I(tτ))h(I)+lnh(I(tτ))h(I))dτ. (4.4)

    For system (2.1), the following Lyapunov functional is considered:

    L = L_1+L_2+L_3.

    From (4.2), (4.3) and (4.4), we can get

    \begin{split} \frac{dL}{dt} = &\mu_1S_1^*(2-\frac{S_1}{S_1^*}-\frac{S_1^*}{S_1}) +(\mu_2-b)S_2^*(3-\frac{S_2}{S_2^*}-\frac{S_1S_2^*}{S_1^*S_2}-\frac{S_1^*}{S_1})\\ &+bS_2^*(2-\frac{S_1S_2^*}{S_1^*S_2}-\frac{S_1^*S_2}{S_1S_2^*})\\ &+nS_2^*f(I^*)(3+\frac{f(I)}{f(I*)}-\frac{S_1^*}{S_1}-\frac{S_1S_2^*}{S_2S_1^*} -\frac{I}{I^*}-\frac{S_2f(I)I^*}{S_2^*f(I^*)I})\\ &+S_2^*g(B^*)(3+\frac{g(B)}{g(B^*)}-\frac{S_1^*}{S_1}-\frac{S_1S_2^*}{S_2S_1^*}-\frac{I}{I^*} -\frac{S_2g(B)I^*}{S_2^*g(B^*)I})\\ &+S_2^*g(B^*)(\frac{h(I)}{h(I^*)}-\frac{B}{B^*}-\frac{B^*h(I)}{Bh(I^*)}+1)\\ &+S_2^*f(I^*)\int_0^{h}\varphi(\tau)\mathcal{F}(\tau)d\tau +\frac{S_2^*g(B^*)}{n}\int_0^{h}\varphi(\tau)\mathcal{H}(\tau)d\tau, \end{split} (4.5)

    where

    \begin{split} \int_0^{h}\varphi(\tau)\mathcal{F}(\tau)d\tau = &\int_0^{h}\varphi(\tau)(\frac{S_2f(I)I^*}{S_2^*f(I^*)I} -\frac{S_2I^*f(I(t-\tau))}{S_2^*f(I^*)I}+\ln\frac{f(I(t-\tau))}{f(I)})d\tau\\ = &\int_0^{h}\varphi(\tau)(\nu(\frac{S_2I^*f(I(t-\tau))}{S_2^*f(I^*)I}) -\nu(\frac{S_2f(I)I^*}{S_2^*f(I^*)I}))d\tau\\ = &\int_0^{h}\varphi(\tau)\nu(\frac{S_2I^*f(I(t-\tau))}{S_2^*f(I^*)I})d\tau -n\nu(\frac{S_2f(I)I^*}{S_2^*f(I^*)I}), \end{split} (4.6)

    and

    \begin{split} \int_0^{h}\varphi(\tau)\mathcal{H}(\tau)d\tau = &\int_0^{h}\varphi(\tau)(\frac{h(I)B^*}{h(I^*)B} -\frac{B^*h(I(t-\tau))}{h(I^*)B}+\ln\frac{h(I(t-\tau))}{h(I)})d\tau\\ = &\int_0^{h}\varphi(\tau)\nu(\frac{B^*h(I(t-\tau))}{h(I^*)B})d\tau -n\nu(\frac{h(I)B^*}{h(I^*)B}). \end{split} (4.7)

    it follows from (4.5), (4.6) and (4.7) that

    \begin{split} \frac{dL}{dt} = &\mu_1S_1^*(2-\frac{S_1}{S_1^*}-\frac{S_1^*}{S_1}) +(\mu_2-b)S_2^*(3-\frac{S_2}{S_2^*}-\frac{S_1S_2^*}{S_1^*S_2}-\frac{S_1^*}{S_1})\\ &+bS_2^*(2-\frac{S_1S_2^*}{S_1^*S_2}-\frac{S_1^*S_2}{S_1S_2^*}) +nS_2^*f(I^*)(\frac{f(I)}{f(I^*)}-1)(1-\frac{f(I^*)I}{f(I)I^*})\\ &+nS_2^*f(I^*)(\nu(\frac{S_1^*}{S_1})+\nu(\frac{S_1S_2^*}{S_1^*S_2}) +\nu(\frac{S_2f(I)I^*}{S_2^*f(I^*)I})+\nu(\frac{f(I^*)I}{f(I)I^*}))\\ &+S_2^*g(B^*)((\frac{g(B)}{g(B^*)}-1)(1-\frac{g(B^*)B}{g(B)B^*}) +(\frac{h(I)}{h(I^*)}-1)(1-\frac{h(I^*)I}{h(I)I^*}))\\ &+S_2^*g(B^*)(\nu(\frac{S_1^*}{S_1})+\nu(\frac{S_1S_2^*}{S_1^*S_2}) +\nu(\frac{S_2g(B)I^*}{S_2^*g(B^*)I})+\nu(\frac{g(B^*)B}{g(B)B^*}))\\ &+S_2^*g(B^*)(\nu(\frac{h(I)B^*}{h(I^*)B})+\nu(\frac{h(I^*)I}{h(I)I^*}))\\ &+S_2^*f(I^*)(\int_0^{h}\varphi(\tau)\nu(\frac{S_2I^*f(I(t-\tau))}{S_2^*f(I^*)I})d\tau -n\nu(\frac{S_2f(I)I^*}{S_2^*f(I^*)I}))\\ &+\frac{S_2^*g(B^*)}{n}(\int_0^{h}\varphi(\tau)\nu(\frac{B^*h(I(t-\tau))}{h(I^*)B})d\tau -n\nu(\frac{h(I)B^*}{h(I^*)B}))\\ \leq &\mu_1S_1^*(2-\frac{S_1}{S_1^*}-\frac{S_1^*}{S_1}) +(\mu_2-b)S_2^*(3-\frac{S_2}{S_2^*}-\frac{S_1S_2^*}{S_1^*S_2}-\frac{S_1^*}{S_1})\\ &+bS_2^*(2-\frac{S_1S_2^*}{S_1^*S_2}-\frac{S_1^*S_2}{S_1S_2^*}) +nS_2^*f(I^*)(\frac{f(I)}{f(I^*)}-1)(1-\frac{f(I^*)I}{f(I)I^*})\\ &+S_2^*g(B^*)((\frac{g(B)}{g(B^*)}-1)(1-\frac{g(B^*)B}{g(B)B^*}) +(\frac{h(I)}{h(I^*)}-1)(1-\frac{h(I^*)I}{h(I)I^*}))\\ \end{split}

    By Lemma 2.1, it can conclude that

    \frac{dL}{dt} = \frac{dL_1}{dt}+\frac{dL_2}{dt}+\frac{dL_3}{dt}\leq 0.

    The equality \frac{dL}{dt} = 0 suggests that \frac{S_1^*}{S_1} = 1 , \frac{S_1S_2^*}{S_1^*S_2} = 1 , \frac{f(I)}{f(I^*)} = 1 and \frac{g(B)}{g(B^*)} = 1 , it implies that E_* is the maximum invariant set of system (2.1) in the set \{\frac{dL}{dt} = 0\} . Using Lemma 4.1, the endemic steady state E_* is globally asymptotically stable.

    By choosing a specific kernel function and some infection functions, system (2.1) can be evolved into different dynamic models with time delays. In this section, some such examples are used to further illustrate theoretical results.

    Example: Consider the kernel function \rho(\tau) = \delta(\tau-\tau_0) , where \delta is the Dirac delta function. Then system (2.1) can be rewritten as

    \begin{equation} \begin{cases} \frac{dS_1}{dt} = A+bS_2-\mu_1S_1-\sigma_1S_1,\\ \frac{dS_2}{dt} = \sigma_1S_1-\alpha S_2 B-\beta S_2f(I(t-\tau_0))e^{-(\mu_2+c)\tau_0}-\mu_2S_2, \\ \frac{dI}{dt} = \alpha S_2 B+\beta S_2f(I(t-\tau_0))e^{-(\mu_2+c)\tau_0}-(\mu_2+c)I,\\ \frac{dB}{dt} = kI(t-\tau_0)e^{-(\mu_2+c)\tau_0}-dB. \end{cases} \end{equation} (5.1)

    Where A, b, \mu_1, \mu_2, \sigma_1, \alpha, \beta, c, k and d > 0 , g(B) = \alpha B and h(I) = kI . If f(I) = \beta I , it is easy to verify that f(I), g(B) and h(I) satisfy the assumptions (H_1)-(H_3) . Therefore, the basic reproduction number of system (5.1) can be defined as

    \begin{equation*} R_0 = \frac{\beta S_2^0e^{-(\mu_2+c)\tau_0}}{\mu_2+c}+\frac{k\alpha S_2^0e^{-(\mu_2+c)\tau_0}}{d(\mu_2+c)}. \end{equation*}

    In this case, if R_0\leq1 , the disease-free equilibrium is globally asymptotically stable and the endemic equilibrium is also globally asymptotically stable if R_0 > 1 (see Figures 1 and 2). As is shown in Figure 2, initial values and time delays can not change dynamic properties of system (2.1) when R_0 is greater than one or less than or equal to one. However, from Figure 2b, it is easy to find that time delay \tau_0 has a significant impact on the positive equilibrium of system (5.1).

    Figure 1.  The time series diagrams of system (5.1) illustrate the effect of the initial values. The values of the parameters are A = 160, b = 0.04, \sigma_1 = 0.2, \mu_1 = 0.01, \beta = 2.1\times 10^{-5} , \alpha = 8.158\times 10^{-6}, \mu_2 = 0.05 , \tau_0 = 1, k = 2, d = 4 . Figure 1a illustrates the time series diagrams when R_0 = 0.6463 (c = 0.3) ; Figure 1b shows the time series diagrams when R_0 = 1 (c = 0.2) .
    Figure 2.  The time series diagrams of system (5.1) illustrate the effect of the initial values and the latency delay \tau_0 . The values of the parameters are A = 160, b = 0.04, \sigma_1 = 0.2 , \mu_1 = 0.01, \beta = 2.1\times 10^{-5} , \alpha = 8.158\times 10^{-6}, \mu_2 = 0.05, c = 0.1 , k = 2, d = 4 . Figure 2a illustrates the time series diagrams when R_0 = 1.8420 (\tau_0 = 1) ; Figure 2b shows the time series diagrams when R_0 = 1.8420(\tau_0 = 1), 1.5854(\tau_0 = 2) and 1.3646(\tau_0 = 3) .

    If f(I) = \beta I^2 [28], then f(I) does not satisfy hypothesis (H_2) , that is, \frac{f(I)}{I} is a monotonically increasing function. From Figure 3, it is easy to see that the system (5.1) appears periodic oscillation behavior under certain conditions.

    Figure 3.  The time series diagrams and phase diagrams of system (5.1) illustrate the effect of the incidence rate f(I) with the initial value (2000, 5000, 2000, 10) . The values of the parameters are A = 330, b = 0.04, \sigma_1 = 0.2 , \mu_1 = 0.01, \beta = 1\times 10^{-9}, \alpha = 8.158\times 10^{-6} , \mu_2 = 0.05, c = 0.02 , k = 2, d = 4, \tau_0 = 1 .

    In this article, a general S_1S_2IB dynamics model with distributed time delay for animal brucellosis is formulated. Under the assumptions of general biological significance, the non-negative and boundedness of the solution of system (2.1) is first proved. And then the global dynamics of the steady-state solution of system (2.1) are analyzed by constructing Lyapunov functional, it is found that the dynamic properties of equilibria depend on the basic reproduction number R_0 : If R_0\leq 1 , animal brucellosis will eventually die out regardless of the initial value; and if R_0 > 1 , the spread of animal brucellosis is persistent and it eventually reaches the endemic steady state. These results imply that distributed time delay does not change the dynamic properties of system (2.1) when R_0 is greater than one or less than or equal to one. Finally, similar to numerical methods in [35,36], the stability results and other dynamic behaviors are further illustrated through numerical simulation, it turns out that the system experiences periodic oscillations if the assumption (H_2) is not satisfied. In other words, the system (2.1) may exhibit more complex dynamic behaviors if the function \frac{f}{I}, \frac{g}{B} or \frac{h}{I} is monotonically increasing. In these cases, the impact of distributed time delay on other dynamical behaviors of system (2.1) is not completely clear, then we leave these for future work.

    This research is partially supported by the National Youth Science Foundation of China (11501528), the National Sciences Foundation of China(11571324) and the Fund for Shanxi "1331KIRT".

    The authors declare that they have no competing interests.



    [1] I. Gusarov, E. Nudler, Protein S-nitrosylation: enzymatically controlled, but intrinsically unstable, post-translational modification, Mol. Cell, 69 (2018), 351-353. doi: 10.1016/j.molcel.2018.01.022
    [2] W. Deng, Y. Wang, L. Ma, Y. Zhang, S. Ullah, Y. Xue, Computational prediction of methylation types of covalently modified lysine and arginine residues in proteins, Briefings Bioinf., 18 (2016), 647-658.
    [3] L. Kiemer, J. D. Bendtsen, N. Blom, NetAcet: prediction of N-terminal acetylation sites, Bioinformatics, 21 (2005), 1269-1270. doi: 10.1093/bioinformatics/bti130
    [4] Y. D. Khan, N. Rasool, W. Hussain, S. A. Khan, K. C. Chou, iPhosT-PseAAC: identify phosphothreonine sites by incorporating sequence statistical moments into PseAAC, Anal. Biochem., 550 (2018), 109-116. doi: 10.1016/j.ab.2018.04.021
    [5] M. M. Hasan, B. Manavalan, M. S. Khatun, H. Kurata, Prediction of S-nitrosylation sites by integrating support vector machines and Random Forest, Mol. Omics, 15 (2019), 451-458. doi: 10.1039/C9MO00098D
    [6] V. Fernando, X. Zheng, Y. Walia, V. Sharma, J. Letson, S. Furuta, S-Nitrosylation: an emerging paradigm of redox signaling, Antioxid. (Basel), 8 (2019), 404. doi: 10.3390/antiox8090404
    [7] H. Hayashi, D. T. Hess, R. Zhang, K. Sugi, H. Gao, B. L. Tan, et al., S-nitrosylation of β-arrestins biases receptor signaling and confers ligand independence, Mol. Cell, 70 (2018), 473-487. doi: 10.1016/j.molcel.2018.03.034
    [8] S. Rizza, S. Cardaci, C. Montagna, G. Di Giacomo, D. De Zio, M. Bordi, et al., S-nitrosylation drives cell senescence and aging in mammals by controlling mitochondrial dynamics and mitophagy, Proc. Natl. Acad. Sci., 115 (2018), 3388-3397.
    [9] F. Li, P. Sonveaux, Z. N. Rabbani, S. Liu, B. Yan, Q. Huang, et al., Regulation of HIF-1α stability through S-nitrosylation, Mol. Cell, 26 (2017), 63-74.
    [10] Z. Wang, Protein S-nitrosylation and cancer, Cancer Lett., 320 (2012), 123-129. doi: 10.1016/j.canlet.2012.03.009
    [11] T. S. Wijasa, M. Sylvester, N. Brocke-Ahmadinejad, S. Schwartz, F. Santarelli, V. Gieselmann, et al., Quantitative proteomics of synaptosome S-nitrosylation in Alzheimer's disease, J. Neurochem., 152 (2020), 710-726.
    [12] M. Piroddi, A. Palmese, F. Pilolli, A. Amoresano, P. Pucci, C. Ronco, F. Galli, Plasma nitroproteome of kidney disease patients, Amino Acids, 40 (2011), 653-667. doi: 10.1007/s00726-010-0693-1
    [13] A. Nott, P. M. Watson, J. D. Robinson, L. Crepaldi, A. Riccio, S-nitrosylation of histone deacetylase 2 induces chromatin remodelling in neurons, Nature, 455 (2008), 411-415. doi: 10.1038/nature07238
    [14] G. Huang, J. Li, C. Zhao, Computational prediction and analysis of associations between small molecules and binding-associated S-nitrosylation sites, Molecules, 23 (2018), 954. doi: 10.3390/molecules23040954
    [15] J. R. Burgoyne, P. Eaton, A rapid approach for the detection, quantification, and discovery of novel sulfenic acid or S-nitrosothiol modified proteins using a biotin-switch method, Methods Enzymol., 473 (2010), 281-303.
    [16] G. Hao, B. Derakhshan, L. Shi, F. Campagne, S. S. Gross, SNOSID, a proteomic method for identification of cysteine S-nitrosylation sites in complex protein mixtures, Proc. Natl. Acad. Sci. U. S. A., 103 (2006), 1012-1017.
    [17] W. R. Qiu, B. Q. Sun, X. Xiao, D. Xu, K. C. Chou, iPhos-PseEvo: identifying human phosphorylated proteins by incorporating evolutionary information into general PseAAC via grey system theory, Mol. Inf., 36 (2017), 1600010. doi: 10.1002/minf.201600010
    [18] W. R. Qiu, A. Xu, Z. C. Xu, C. H. Zhang, X. Xiao, Identifying acetylation protein by fusing its PseAAC and functional domain annotation, Front. Bioeng. Biotechnol., 7 (2019), 311. doi: 10.3389/fbioe.2019.00311
    [19] Y. Xue, Z. Liu, X. Gao, C. Jin, L. Wen, X. Yao, J. Ren, GPS-SNO: computational prediction of protein S-nitrosylation sites with a modified GPS algorithm, PLoS One, 5 (2010), e11290. doi: 10.1371/journal.pone.0011290
    [20] T. Y. Lee, Y. J. Chen, T. C. Lu, H. D. Huang, Y. J. Chen, SNOSite: exploiting maximal dependence decomposition to identify cysteine S-nitrosylation with substrate site specificity, PLoS One, 6 (2011), e21849. doi: 10.1371/journal.pone.0021849
    [21] Y. Xu, J. Ding, L. Y. Wu, K. C. Chou, iSNO-PseAAC: predict cysteine S-nitrosylation sites in proteins by incorporating position specific amino acid propensity into pseudo amino acid composition, PLoS One, 8 (2010), e55844.
    [22] A. Siraj, T. Chantsalnyam, H. Tayara, K. T. Chong, RecSNO: prediction of protein S-nitrosylation sites using a recurrent neural network, IEEE Access, 9 (2021), 6674-6682. doi: 10.1109/ACCESS.2021.3096125
    [23] X. Cheng, X. Xiao, K. C. Chou, pLoc_bal-mGneg: predict subcellular localization of gram-negative bacterial proteins by quasi-balancing training dataset and general PseAAC, J. Theor. Biol., 458 (2018), 92-102. doi: 10.1016/j.jtbi.2018.09.005
    [24] K. C. Chou, Some remarks on protein attribute prediction and pseudo amino acid composition, J. Theor. Biol., 273 (2011), 236-247. doi: 10.1016/j.jtbi.2010.12.024
    [25] Y. Z. Chen, Y. R. Tang, Z. Y. Sheng, Z. Zhang, Prediction of mucin-type O-glycosylation sites in mammalian proteins using the composition of k-spaced amino acid pairs, BMC Bioinf., 9 (2008), 101. doi: 10.1186/1471-2105-9-101
    [26] F. N. Auliah, A. N. Nilamyani, W. Shoombuatong, M. A. Alam, M. M. Hasan, H. Kurata, PUP-fuse: prediction of protein pupylation sites by integrating multiple sequence representations, Int. J. Mol. Sci., 22 (2021), 2120. doi: 10.3390/ijms22042120
    [27] Y. Liu, Z. Yu, C. Chen, Y. Han, B. Yu, Prediction of protein crotonylation sites through LightGBM classifier based on SMOTE and elastic net, Anal. Biochem., 609 (2020), 113903. doi: 10.1016/j.ab.2020.113903
    [28] Y. Xie, X. Luo, Y. Li, L. Chen, W. Ma, J. Huang, et al., DeepNitro: prediction of protein nitration and nitrosylation sites by deep learning, Genomics, Proteomics Bioinf., 16 (2018), 294-306. doi: 10.1016/j.gpb.2018.04.007
    [29] W. Qiu, Z. Lv, Y. Hong, J. Jia, X. Xiao, BOW-GBDT: a GBDT classifier combining with artificial neural network for identifying GPCR-drug interaction based on wordbook learning from sequences, Front. Cell Dev. Biol., 8 (2021), 623858. doi: 10.3389/fcell.2020.623858
    [30] S. Kawashima, M. Kanehisa, AAindex: amino acid index database, Nucleic Acids Res., 28 (2000), 374. doi: 10.1093/nar/28.1.374
    [31] N. V. Chawla, K. W. Bowyer, L. O. Hall, W. P. Kegelmeyer, SMOTE: synthetic minority over-sampling technique, J. Artif. Intell. Res., 16 (2002), 321-357. doi: 10.1613/jair.953
    [32] M. J. Kim, D. K. Kang, H. B. Kim, Geometric mean based boosting algorithm with over-sampling to resolve data imbalance problem for bankruptcy prediction, Expert Syst. Appl., 42 (2015), 1074-1082. doi: 10.1016/j.eswa.2014.08.025
    [33] S. Yan, W. Qian, Y. Guan, B. Zheng, Improving lung cancer prognosis assessment by incorporating synthetic minority oversampling technique and score fusion method, Med. Phys., 43 (2016), 2694-2703. doi: 10.1118/1.4948499
    [34] N. Habib, M. M. Hasan, M. M. Reza, M. M. Rahman, Ensemble of CheXNet and VGG-19 feature extractor with Random Forest classifier for pediatric pneumonia detection, SN Comput. Sci., 1 (2020), 359. doi: 10.1007/s42979-020-00373-y
    [35] K. Ghosh, A. Sarkar, A. Banerjee, S. Chatterjee. Performance improvement of convolutional neural network using random under sampling, in Advances in Smart Communication Technology and Information Processing, Springer, (2021), 207-217.
    [36] S. Hua, Z. Sun, Support vector machine approach for protein subcellular localization prediction, Bioinformatics, 17 (2001), 721-728. doi: 10.1093/bioinformatics/17.8.721
    [37] H. Bian, M. Guo, J. Wang, Recognition of mitochondrial proteins in plasmodium based on the tripeptide composition, Front. Cell Dev. Biol., 8 (2020), 578901. doi: 10.3389/fcell.2020.578901
    [38] H. Zou, T. Hastie, Regularization and variable selection via the elastic net, J. R. Stat. Soc. Ser. B (Stat. Methodol.), 67 (2005), 301-320. doi: 10.1111/j.1467-9868.2005.00503.x
    [39] G. Chen, M. Cao, J. Yu, X. Guo, S. Shi, Prediction and functional analysis of prokaryote lysine acetylation site by incorporating six types of features into Chou's general PseAAC, J. Theor. Biol., 461 (2019), 92-101. doi: 10.1016/j.jtbi.2018.10.047
    [40] W. R. Qiu, B. Q. Sun, X. Xiao, Z. C. Xu, J. H. Jia, K. C. Chou, iKcr-PseEns: identify lysine crotonylation sites in histone proteins with pseudo components and ensemble classifier, Genomics, 110 (2018), 239-246. doi: 10.1016/j.ygeno.2017.10.008
    [41] L. Breiman, Random Forests. Mach. Learn., 45 (2001), 5-32. doi: 10.1023/A:1010933404324
    [42] H. Talebi, L. J. M. Peeters, A. Otto, R. Tolosana-Delgado, A truly spatial Random Forests algorithm for geoscience data analysis and modelling, Math. Geosci., 2021.
    [43] Z. Qiu, Q. Liu, Protein-protein interaction site prediction using random forest proximity distance, J. Bioinf. Comput. Biol., 19 (2021), 2050042. doi: 10.1142/S0219720020500420
    [44] S. Cabras, M. E. Castellanos, E. Staffetti, A Random Forest application to contact-state classification for robot programming by human demonstration, Appl. Stochastic Models Bus. Ind., 32 (2016), 209-227. doi: 10.1002/asmb.2145
    [45] N. Friedman, D. Geiger, M. Goldszmidt, Bayesian network classifiers, Mach. Learn., 29 (1997), 131-163. doi: 10.1023/A:1007465528199
    [46] R. Queiroz, T. Berger, K. Czarnecki, Towards predicting feature defects in software product lines, in Proceedings of the 7th International Workshop on Feature-Oriented Software Development, Amsterdam, Netherlands, Association for Computing Machinery, (2016), 58-62.
    [47] H. Bohra, A. Arora, P. Gaikwad, R. Bhand, M. R. Patil, Health prediction and medical diagnosis using Naive Bayes, Ijarcce, 6 (2017), 32-35.
    [48] M. S. Ahmed, M. Shahjaman, E. Kabir, M. Kamruzzaman, Prediction of protein acetylation sites using kernel Naive Bayes classifier based on protein sequences profiling, Bioinformation, 14 (2018), 213-218. doi: 10.6026/97320630014213
    [49] F. Nigsch, A. Bender, B. van Buuren, J. Tissen, E. Nigsch, J. B. Mitchell, Melting point prediction employing K-Nearest Neighbor algorithms and genetic parameter optimization, J. Chem. Inf. Model., 46 (2006), 2412-2422. doi: 10.1021/ci060149f
    [50] A. Wirdiani, P. Hridayami, A. Widiari, K. Rismawan, P. Candradinata, I. Jayantha, Face identification based on K-Nearest Neighbor, Sci. J. Inf., 6 (2019), 150-159.
    [51] O. Borgohain, M. Dasgupta, P. Kumar, G. Talukdar, Performance analysis of nearest neighbor, K-Nearest Neighbor and weighted K-Nearest Neighbor for the cassification of Alzheimer disease, in Soft Computing Techniques and Applications, S. Borah, R. Pradhan, N. Dey, P. Gupta, Ed., Springer, Singapore, (2021), 295-304.
    [52] S. Huang, M. Huang, Y. Lyu, A novel approach for sand liquefaction prediction via local mean-based pseudo nearest neighbor algorithm and its engineering application, Adv. Eng. Inf., 41 (2019), 100918. doi: 10.1016/j.aei.2019.04.008
    [53] T. Chen, C. Guestrin, XGBoost: a scalable tree boosting system, in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, California, USA, Association for Computing Machinery, (2016), 785-794.
    [54] I. Babajide Mustapha, F. Saeed, Bioactive molecule prediction using extreme gradient boosting, Molecules, 21 (2016), 983. doi: 10.3390/molecules21080983
    [55] J. H. Friedman, Greedy function approximation: a gradient boosting machine, Ann. Stat., 29 (2001), 1189-1232. doi: 10.1214/aos/1013203450
  • Reader Comments
  • © 2021 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(3140) PDF downloads(77) Cited by(0)

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog