Gene transcription results in the formation of a specific cellular response to the signal

Fig. 2. Information exchange in the cellular environment

Gene transcription results in the formation of a specific cellular response to the signal

Fig. 2. Information exchange in the cellular environment such a signal transduction pathway is mostly gene transcription, other possibilities are the reorganization of intracellular structure such as the cell cytoskeleton or the internalization and externalization in and out of the cell. Gene transcription means that the cell respond to incoming the signal by production of other factors which are then secreted (transported out of the cell), where it can induce signaling processes in the cell's direct environment. This process is depicted in a simplified manner in figure 2. A cell is shown with a single receptor that is able to receive a very specific signal, i.e. a protein, and to activate a signaling cascade which finally forms the cellular response.

This is the key to information processing. It depends on the type of the signal and the state of the cells (which receptors have been built and which of them are already occupied by particular proteins). Finally, a specific cellular response is induced: either the local state is manipulated and/or a new messaging protein is created. The remote information exchange works analogue. Proteins, peptides, and steroids are used as information particles (hormones) between cells. A signal is released into the blood stream, the medium that carries it to distant cells and induces an answer in these cells which then passes on the information or can activate helper cells (e.g. the Renin-Angiotensin-Aldosteron system [19] and the immune system). The interesting property of this transmission is that the information itself addresses the destination. During differentiation a cell is programmed to express a subset of receptor in order to fulfill a specific function in the tissue. In consequence, hormones in the bloodstream affect only those cells expressing the correct receptor. This is the main reason for the specificity of cellular signal transduction. Of course, cells also express a variety of receptors which regulate the cellular metabolism, survival, and death.

The lessons to learn from biology are the efficient and, above all, the very specific response to a problem, the shortening of information pathways, and the possibility of directing each problem to the adequate helper component. Therefore, the adaptation of mechanisms from cell and molecular biology promises to enable a more efficient information exchange. Besides all the encouraging properties, bio-inspired techniques must be used carefully by modeling biological and technical systems and choosing only adequate solutions.

So, how to use the described methods to WSN and SANET operation and control? The biological model needs to be checked and - partially - adapted

Fig. 3. Architecture and behavior of a local node

to match the tasks in sensor/actuator networks. In the following section, we describe and discuss a solution for network-centric operation and control based on the described biological mechanisms.

3 Rule-Based State Machine for Localized Actuation Control

As already mentioned, three basic mechanisms are used to achieve the demanded goals:

- Each message carries all necessary information to allow the specific handling of the associated data.

- A rule-based programming scheme is used to describe specific actions to be taken after the reception of particular information fragments.

- We do not intend to control the overall system but focus on the operation of the individual node instead (see discussion on emergent system behavior in section 4). We designed simple state machines that control each node whether sensor or actuator.

The complete scheme as adapted from cellular behavior is shown in figure 3. Even though the principles are described later, the general architecture and the behavior can be shortly explained. Depicted is a network node that has four directly connected neighbors (A, B, C, D). The local behavior is controlled by a state machine (n, a) and a set of rules (RuleDB). In this example, a data message of type x is received and transformed locally into a message of type y. Finally, this message is distributed to all neighbors. (Remark: we consider wireless communication. Therefore, each message that a node sends is basically a broadcast to all neighboring nodes.)

3.1 Data-Centric Operation

Classically, communication in ad hoc networks is based on topology information, i.e. routing paths that have been set-up prior to any data exchange. Additionally, each node carries a unique address that is used to distinguish the desired destination. We follow the approach used in typical data-centric communication schemes, e.g. directed diffusion [10], and replace topology information and addressing by data-centric operation. Each message is encoded as follows:

M:={type, region, confidence, content}

Using this description, we can encode measurement data as well as actuator information (type and content). Additionally, the region is included to distinguish messages from the local neighborhood from those that traveled over a long distance. Finally, the confidence value is used to evaluate the message in terms of importance or priority. Measures with a high confidence will have a stronger impact on calculations that those with a lower confidence. The confidence can be changed using aggregation schemes, i.e. two measures of the same value in the same region will lead to a higher confidence.

The following examples demonstrate the capabilities of the message encoding for data-centric operation:

— {temperatureC, [10,20], 0.6, 20} :: A temperature of 20C was measured at the coordinates [10,20]. The confidence is 0.6, therefore, a low-quality sensor was employed.

— {pictureJPG, [10,30], 0.9, "binary JPEG"} :: A picture was taken in format JPEG at the coordinates [10,30].

3.2 Specific Reaction on Received Data

An extensible and flexible rule system is used to evaluate received messages and to provide the "programming" that specifies the cellular response. Even though the message handling in biological cells is more sophisticated, the basic principles including the processing instructions (the DNA) are modeled. Each rule consists of two parts: a number of input values and some output: INPUT ^ OUTPUT. Therefore, typical rules could look like that:

— A A B ^ C :: if both messages A and B were received, a message C is created

Using all the other information available in each message, more complex rules can be derived:

— A(content> 10) ^ A(confidence:= 0.9) :: if the measured value was larger than 10, a copy of A is created with confidence set to 0.9

— A(content= x) A A(content= y) ^ A(content:= x + y) :: two messages of type A are aggregated to a single one by adding their values

Again, an example is provided to reflect the capabilities of the data-centric operation:

— temperatureC(content> 85) ^alarmFire(confidence:= 0.8)

measure receive

transmit actuate

Fig. 4. Simple state machines for sensors (S) and actuators (A)

r input

Rule interpreter output


Fig. 5. Rule interpreter with system input and output

3.3 Simple Local Behavior Control

The local behavior is controlled by simple state machines acting as sensors or actuators. Additionally, an interpreter is checking the installed rules to previously received messages. It uses a queuing subsystem that acts as a generic receptor for all messages and keeps them for a given time. This time control is necessary to prevent queue overflows due to received messages of unknown type. The basic state machines for sensing and transmitting data and receiving and acting on data for sensors and actuators, respectively, are shown in figure 4.

The rule interpreter and its queuing system are depicted in figure 5. Basically, this is the standard behavior of each communication system. Received messages are stored in a local database. After a given timeout, each message is dropped in order to keep the size of the database below a given threshold. Periodically, the rule interpreter compares all received messages against the programmed rule set. A matching rule terminates the search and the rule is applied.

3.4 Case Studies

Two case studies are provided in this section to elaborate the principles and the flexibility of the proposed network-centric operation and control method for sensor/actuator networks: first, data aggregation and emergency calls, and secondly, in-network actuation control. Both examples were also chosen in order to show the benefits of our approach compared to traditional WSN mechanisms.

Data aggregation and emergency calls. We consider a typical scenario for wireless sensor networks. Sensor nodes are distributed over a given area. All nodes are equipped with sensors measuring a particular physical phenomenon, e.g. the temperature. In order to obtain information about the territory, the measurement results are transported to a given sink that analyzes the received temperature information. Additionally, measures exceeding a given threshold represent emergency situations that must be handled separately. In both examples, we assume a priority-based message forwarding scheme on the network layer.

1. Data types

Mtemp := {temperature, position, content, priority} Maiarm := {alarm, position, content, priority}


2. Rule set Aggregation:

Atemp (content)eqBtemp(content) ^ Ctemp(priority := priority a + (1 -


Atemp(content > 70) ^ Baiarm(priority := max(priority, 0.8)

The aggregation rule combines multiple messages containing the same measurement results into a single message. Such aggregated messages must be handles with more care in the network since a packet loss of an aggregate of n messages can be compared to n separate lost packets without aggregation. In our example, the priority of the aggregated message is increased in order to enable the network layer to handle this packet specifically. The emergency rule creates new alarm packets if measurements above 70 degrees were observed. Additionally, the priority is explicitly set to a high value representing the importance of such a message.

3. Evaluation

The benefits of the aggregation and emergency example can be shown easily. Consider the following scenario. All sensor nodes are directly connected to a central base station. In this case, each message must travel exactly one hop before processing. There is no possibility for aggregation to take place. In every other case, multi hop communication is involved and multiple messages can be aggregated. Compared to a pure central processing, the approach always leads to a noticeable reduction of the network load.

In-network actuation control. A second example includes additional actuators. Based on the temperature measurement as discussed before, temperature control should be performed, e.g. by using AC or heatings. Such actuators are controlled by special control messages. In the following description, only the differences and additions to the previous example are shown.

1. Data types

Mcontroi := {control, position, delta, priority}


2. Rule set

Controli: Atemp(content\ = 20) ^ Bcontroi(delta := 20 — content a) Control2: Acontrol ^execute actuation command

Two types of control rules exist. The first one (Controli) can be seen as the in-network processing part. Received messages are verified whether actuation control should take place. In our example, a mean temperature of 20 degrees should be maintained. The second rule (Control2) depicts the actuation initiation. After receiving a control message at an actuator, it performs the necessary actuation as encoded in the message (after checking if it can provide the needed service).

3. Evaluation

Similarly to the previous example, a fully connected network (all sensors and actuators have a direct connection to the base station) will always perform optimal in terms of network overhead. Nevertheless, such a topology is unrealistic considering larger areas to be observed and maintained. In this case, each sensor message must traverse a multi hop path toward the base station. Then, after a meaningful evaluation, the actuation control must travel back to the actuators.

In this case study, another possible topology can be imagined that also leads to a non-optimal operation of the proposed solution: if all sensors build a separate network partition as well as all actuators, then the base station will become the gateway between both networks. In this case all messages must traverse the base, i.e. there is no overhead in terms of duplicate network utilization for sensor and actuation control messages. Admittedly, this scenario is unrealistic as well.

In conclusion, our solution for network-centric operation and control will perform at least as good as a central base station approach and outperform it in most realistic network scenarios, i.e. in networks consisting of a mixture of sensors and actuators.

4 Discussion

Based on the previously stated key requirements, the benefits of the proposed solution are reviewed in the following. Additionally, potential disadvantages or problems are stated and discussed:

— Self-organized operation without central control - The presented approach is based on locally available information only. Using the flexible rule system, arbitrary data-centric operations can be defined enabling the systems to specifically act on each received message.

— Allowance for centralized "helpers" and self-learning properties - Rules can be specified to forward all unknown messages to a central "helper". This system can examine the message, create according rules, and submit these rules to replace/enhance the rules installed in the SANET nodes. Therefore, our method provides at least limited control in a system showing an emergent behavior.

— Reduced network utilization - The network utilization no longer depends on the amount of measurement data to be transmitted to a base station. Instead, the rule system is responsible if and how messages have to be forwarded to more distant regions of the network.

— Accelerated response / actuation - The response time is much smaller than in the centralized approach due to the shortened data paths from measurement to processing, which takes place directly within the network, and the actuation. Depending on the installed rules and their spatial distribution, even boundaries for the response time can be derived.

Potential problems can appear through the inherent characteristics of such self-organizing processes [20], i.e. issues such as predictability of an action, reliability of the communication, and boundaries for response times must be considered. In general, there is no global state information available. Therefore, optimal solutions for the entire network cannot be calculated based on all theoretically available measures. Nevertheless, depending on the rule set, solutions can be derived that approximate the globally optimal solution quite well. Another issue is the necessary pre-programming of the rule sets into all the nodes. If new algorithms should be deployed, which is easy and straightforward using a central control, all or at least many of the distributed nodes must be changed. Fortunately, there are already network-based reprogramming techniques [21] and robot-assisted solutions [22] available to provide this functionality.

5 Conclusion

In this paper, we presented and discussed a methodology for network-centric operation and control of sensor/actuator networks. Inspired by biological information processing, we developed three easy to handle building blocks: data-centric communication, a state machine, and a rule-based decision process. Using these algorithms, the handling and processing of sensor data within the network itself becomes possible. In particular, we demonstrated that a collaborative sensing and processing approach for sensor/actuator networks based on local intelligence is possible. The interaction and collaboration between these nodes finally leads to an optimized system behavior in an emergent way.

Further work is needed in two directions: first, a detailed performance analysis for different application scenarios is necessary in order to rate the practical usability of the approach depending on the scenario. Secondly, it might be helpful if the rule sets are not " programmed" into each node but exchanged and updated on-demand by the nodes themselves in terms of a learning process.


The conducted research on bio-inspired networking is collaborative work with Dr. B. Kruger, Dept. of Physiology, University of Erlangen-Nuremberg, Germany.


1. Estrin, D., Culler, D., Pister, K., Sukhatme, G.S.: Connecting the Physical World with Pervasive Networks. IEEE Pervasive Computing 1, 59-69 (2002)

2. Akyildiz, I.F., Su, W., Sankarasubramaniam, Y., Cayirci, E.: A Survey on Sensor Networks. IEEE Communications Magazine 40, 102-116 (2002)

3. Akyildiz, I.F., Kasimoglu, I.H.: Wireless Sensor and Actor Networks: Research Challenges. Elsevier Ad Hoc Network Journal 2, 351-367 (2004)

4. Margi, C.: A Survey on Networking, Sensor Processing and System Aspects of Sensor Networks. Report, University of California, Santa Cruz (2003)

5. Culler, D., Estrin, D., Srivastava, M.B.: Overview of Sensor Networks. Computer 37, 41-49 (2004)

6. Chong, C.-Y., Kumar, S.P.: Sensor Networks: Evolution, Opportunities, and Challenges. Proceedings of the IEEE 91, 1247-1256 (2003)

7. Parker, L.E.: Current Research in Multi-Robot System. Journal of Artificial Life and Robotics 7 (2004)

8. Kumar, V., Rus, D., Singh, S.: Robot and Sensor Networks for First Responders. IEEE Pervasive Computing 3, 24-33 (2004)

9. Mainwaring, A., Polastre, J., Szewczyk, R., Culler, D., Anderson, J.: Wireless Sensor Networks for Habitat Monitoring. In: First ACM Workshop on Wireless Sensor Networks and Applications, Atlanta, GA, USA, ACM Press, New York (2002)

10. Intanagonwiwat, C., Govindan, R., Estrin, D.: Directed diffusion: A scalable and robust communication paradigm for sensor networks. In: 6th Annual ACM/IEEE International Conference on Mobile Computing and Networking (MobiC0M'00), Boston, MA, USA, pp. 56-67. IEEE Computer Society Press, Los Alamitos (2000)

11. Batalin, M.A., Sukhatme, G.S.: Using a Sensor Network for Distributed MultiRobot Task Allocation. In: IEEE International Conference on Robotics and Automation, New Orleans, LA, USA, May 2003, pp. 158-164. IEEE Computer Society Press, Los Alamitos (2003)

12. Muscettola, N., Nayak, P.P., Pell, B., Williams, B.C.: Remote Agent: To Boldly Go Where No AI System Has Gone Before. Artificial Intelligence 100, 5-48 (1998)

13. Eigen, M., Schuster, P.: The Hypercycle: A Principle of Natural Self Organization. Springer, Berlin (1979)

14. Dressler, F.: Efficient and Scalable Communication in Autonomous Networking using Bio-inspired Mechanisms - An Overview. Informatica - An International Journal of Computing and Informatics 29, 183-188 (2005)

15. Alberts, B., Bray, D., Lewis, J., Raff, M., Roberts, K., Watson, J.D.: Molecular Biology of the Cell, 3rd edn. Garland Publishing, Inc. (1994)

16. Pawson, T.: Protein modules and signalling networks. Nature 373, 573-580 (1995)

17. Dressler, F., KrUger, B.: Cell biology as a key to computer networking. In: German Conference on Bioinformatics 2004 (GCB'04), Poster Session, Bielefeld, Germany (2004)

18. KrUger, B., Dressler, F.: Molecular Processes as a Basis for Autonomous Networking. IPSI Transactions on Advances Research: Issues in Computer Science and Engineering 1, 43-50 (2005)

19. Guyton, A.: Blood pressure control - special role of the kidneys and body fluids. Science 252, 1813-1816 (1991)

20. Prehofer, C., Bettstetter, C.: Self-Organization in Communication Networks: Principles and Design Paradigms. IEEE Communications Magazine 43, 78-85 (2005)

21. Jeong, J., Culler, D.: Incremental Network Programming for Wireless Sensors. In: First IEEE International Conference on Sensor and Ad hoc Communications and Networks (IEEE SECON), IEEE Computer Society Press, Los Alamitos (2004)

22. Fuchs, G., Truchat, S., Dressler, F.: Distributed Software Management in Sensor Networks using Profiling Techniques. In: 1st IEEE/ACM International Conference on Communication System Software and Middleware (IEEE COMSWARE 2006): 1st International Workshop on Software for Sensor Networks (SensorWare 2006), New Dehli, India, ACM Press, New York (2006)

A Computationally Fast and Parametric Model to Estimate Protein-Ligand Docking Time for Stochastic Event Based Simulation

Preetam Ghosh, Samik Ghosh, Kalyan Basu, and Sajal K. Das

Biological Networks (BONE) Research Group, Dept. of Comp. Sc. & Engg. The University of Texas at Arlington, TX, USA {ghosh, sghosh, basu, das}@cse.uta.edu

Abstract. This paper presents a computationally fast analytical model to estimate the time taken for protein-ligand docking in biological pathways. The environment inside the cell has been reported to be unstable with a considerable degree of randomness creating a stochastic resonance. To facilitate the understanding of the dynamic behavior of biological systems, we propose an "in silico" stochastic event based simulation. The implementation of this simulation requires the computation of the execution times of different biological events such as the protein-ligand docking process (time required for ligand-protein binding) as a random, variable. The next event time of the system is computed by adding the event execution time to the clock value of the event start time. Our mathematical model takes special consideration of the actual biological process of ligand-protein docking with emphasis on the structural configurations of the ligands, proteins and the binding mechanism that enable us to control the model parameters considerably. We use a modification of the collision theory based approach to capture the randomness of this problem in discrete time and estimate the first two moments of this process. The numerical results for the first moment show promising correspondence with experimental results and demonstrate the efficacy of our model.

1 Introduction

The Genome project [1], tremendous advancement in micro-array analysis techniques [2], and large scale assay technologies like cDNA array [3] are generating large volumes of scientific data for biological systems, from microbes to homosapiens. We are now in an era where our capability of generating relevant data is less of an obstacle than our understanding of biological systems or networks. The system simulation of biological processes is now considered an important technique to understand its dynamics. The concept of "in silico" [4,5,6] or discrete event based modeling has been successfully applied to study many complex systems. Our goal is to build a similar discrete event based framework for complex biological systems [9,12]. Our main motivation is to overcome the complexities of current mesoscale and stochastic simulation methods and create a flexible simulation

C. Priami (Ed.): Trans. on Comput. Syst. Biol. VIII, LNBI 4780, pp. 14-41, 2007. © Springer-Verlag Berlin Heidelberg 2007

framework. The mesoscale model deals with rate equation based kinetic models and uses continuous time deterministic techniques. Such model is closely related to a rate constant derived from measurements, which captures the experimental boundary conditions and physical reaction dynamics. This model solves complex differential equations corresponding to chemical reactions using numerical integration. Numerical integrations are normally computed at 10~6 time steps i.e., the instruction set for each equation is computed every microsecond. The current super scalar computers with dual processors can compute around 2 — 3 machine instructions per clock cycle. Thus the maximum number of instructions that can be supported per microsecond is about 2000 — 3000 with a 1 GHZ clock. This might not be sufficient to solve a system of even 1000 equations (25 machine instructions per equation results in a > 10 times speed reduction). Since a biological system involves a very large number of differential equations (> 1000), the mesoscale model is not suitable for a large system. Also the stochastic resonance [13] specially for protein creation and other signaling pathways are not properly captured in the mesoscale model unless it is modified to the stochastic mode. The Stochastic simulation models are based on rate equations, namely Gillespie technique [14] and its variations such as Kitano's Cell Designer [15], DARPA's BioSpice [16], StochSim [17], Cell Illustrator [18] etc. and it has more computational overhead due to the random number computations at each time step. Due to the large number of protein complexes in a cell, these models lead to combinatorial explosion in the number of reactions, thus making them unmanageable for complex signaling pathway problems. These limitations of current techniques and the potential opportunity to integrate multi-layer events under one simulation framework motivates our work.

Fig 1 presents an overview of our multi-scale discrete event based framework. We define a biological network/system as a collection of cells which are in turn a collection of biological processes. Each process comprises a number of functions, where a function will be modeled as an event. A unique pathway will be defined as a biological process consisting of a number of biological functions. The fundamental entity in our proposed mathematical model is an "event" which represents a biological function with relevant boundary conditions. These event models are then used to develop a stochastic discrete-event simulation. The interactions between cells are captured in the Biological network view. Then for every cell all the biological processes are identified and their relationship is defined in the cell view of the biological system. Each pathway is described by the event diagram of the biological process. There is a considerable amount of pathway information currently captured in different bioinformatics databases [17],[37],[38]. Thus the completion of these event diagrams seem feasible now. All these events are statistically modeled using the functionality of that particular biological event. The mathematical abstraction of these events can be selected based on the complexity of the event dynamics and its mechanism. This flexibility of using different mathematical abstractions for different types of events make this technique more attractive than ^-calculus [41,42,43,44] or other types of stochastic system modeling. This paper focuses on the modeling of one such event: 'ligand-protein'

Fig. 1. Multi-scale discrete event model framework docking. We present a parametric mathematical model to compute the execution time (or holding time) for ligand-protein binding (i.e., time required for the binding to occur) which is also computationally fast. We have already developed a few models for modeling other biological events like (1) cytoplasmic reactions [10] [11], (2) Protein DNA binding [40] and (3) arrival of Mg2+ molecule signal from external cell environment to trigger a pathway [39].

1.1 Related Works

Most of the work on protein-ligand docking use Brownian dynamic simulations to model the mechanism. From the point of view of kinetics, protein docking should entail distinct kinetic regimes where different driving forces govern the binding process at different times [26,27,28]. This is because of the free energy funnel created by the binding site of the protein. The funnel distinguishes three kinetic regimes. First, nonspecific diffusion (regime I) brings the molecules to close proximity. This is the motion created by the random collision of the molecules. Second, in the recognition stage (regime II), the chemical affinity steers the molecules into relatively well oriented encounter complexes (« 5 x 10"10 m), overcoming the mostly entropic barrier to binding. Brownian dynamics simulation of this regime [19] were also found to be consistent with a significant narrowing of the binding pathway to the final bound conformation. Finally, regime III corresponds to the docking stage where short-range forces mold the high affinity interface of the complex structure.

Long-range electrostatic effects can heavily bias the approach of the molecules to favor reactive conditions. This effect was shown to be important for many association processes, including those of proteins with DNA [8], proteins with highly charged small molecules [29], and proteins with oppositely charged protein substrates [30,31,32,33,34]. These systems have been thoroughly studied, and are frequently regarded as typical examples of binding phenomena. Electrostatics is clearly not the only force that can affect the association rate. In addition to electrostatics, the most important process contributing to the binding free energy is desolvation, i.e., the removal of solvent both from nonpolar (hydrophobic) and polar atoms [35]. It is generally accepted that partial desol-vation is always a significant contribution to the free energy in protein-protein association, and it becomes dominant for complexes in which the long-range electrostatic interactions are weak [36]. Brownian dynamics simulations to study the effects of desolvation on the rates of diffusion-limited protein-protein association have been reported in [19].

In this paper, our goal is to introduce a collision theory model to explain the temporal kinetics of ligand-protein docking. This is a simplified model which does not incorporate the effects of electrostatic forces and desolvation directly as parameters of the model but consider their effects through the random molecular motion of the proteins in the binding environment. This simplification of the model makes it a random collision problem within the cell and gives us a fairly accurate but computationally fast model for the docking time estimate to be used by our stochastic simulator. Note that the Gillespie simulator considers the docking process as another rate-based equation (a measured quantity that encapsulates all the kinetic properties of the process during the experiment), whereas our proposed model can incorporate the salient features of the docking process along with the structural and functional properties of the protein-ligand pair. This parametric presentation of the binding process makes the model generic in nature and can be easily used for other cases of protein-ligand binding where the assumptions are valid. The results generated by this model are very close to experimental estimates. The main conclusion of our work is that the total time required for docking is mostly contributed by the repeated collisions of the ligand with the protein. Also because the ligand on arriving inside the cell compartment spends most of the time (for binding) away from the protein (to which it binds), the effects of electrostatic force and desolvation are negligible in the binding time estimation. However, they play a significant role in the determination of the free energy change of the docked complex [19] which in turn is used in determining the probability of docking as stated later in the paper.

2 Analytical Model: Ligand-Protein Docking

Let us consider the docking between a protein A and a ligand B. Let the total number of surface binding points in A be ua and that in B be nB. The number of surface docking points to produce the AB complex is denoted by ns, such that:

ns docking points

Binding site of A

Protein A

Fig. 2. The protein docking mechanism

We assume that the ns docking points are all contiguous. We also assume that if any three of the docking points is hit by the ligand during a collision, the attractive force of the amino acid side-chain will force the ligand to change orientation so that it can bind to the site. This assumption has a few limitations which we will discuss in Section 5. Now, let the total probability of hitting the site during a collision for successful docking be pf. The probability of hitting the binding site at only one of the docking points is pJ, = jttjj^t^- Similarly, the probability of hitting the binding site at i docking points is given by:

Also, let pb denote the probability that the ligand collides with the protein A with sufficient kinetic energy for successful docking. Hence, the total probability that the ligand hits the binding site while colliding with the protein, pt, is given by:

In general, the process of protein-ligand association can be described by a three-step reaction mechanism as follows:

0 0

Post a comment