KR102269932B1 - Method for identifying drug target molecules based on influence analysis of molecules in signaling networks - Google Patents

Abstract

A method of determining a control target node to be controlled for controlling an output node of a signaling network is disclosed. In this method, the influence of each of a plurality of nodes included in the first signaling network is calculated for a first output node to calculate a plurality of influences corresponding to each of the plurality of nodes. a calculation step, and a control for controlling the first output node by selecting some of the plurality of nodes according to a control target preset for the first output node based on the values of the plurality of influences and a control target node determining step of determining as a target node. In this case, the influence of any first node among the plurality of nodes on the first output node indicates the degree of influence of the change in the activity of the first node on the change in the activity of the first output node.

Description

Method for identifying drug target molecules based on influence analysis of molecules in signaling networks

The present invention relates to information processing technology using a computer, and more particularly, to a technology for analyzing information about the signal flow of a signal transduction network modeled by signal transduction between molecules in a cell.

The present invention can be practiced using the novel concepts presented by the inventor. Accordingly, the new concepts described above will be described below. Not all content described in this chapter should be treated as prior art.

1 shows the concept of a biochemical reaction and a directed signed link in the signaling network dealt with in the present invention.

As used herein, the 'signal transduction network' refers to a network composed of interactions between various signal transduction molecules. A single signaling network may include a plurality of links. In this case, each link may be the directional code link having a direction. Therefore, two selected nodes (=signaling molecules) in the signaling network can be connected by a directed code link. Figure 2 schematically presents an example of the structure of a complex signaling network.

When a specific protein or compound binds to a cell membrane receptor outside the cell, or the function of a specific signaling molecule is affected by a gene mutation or a drug inside the cell, a series of biochemical reactions occur, resulting in important physiological functions of the cell. function will change. A series of biochemical reactions occurring at this time may be referred to as signal transduction (signaling).

The biochemical reaction shown in FIG. 1 (a) is simply the activation of a protein by a phosphate group-bonding catalyst, but when such a biochemical reaction occurs in a chain and an important function of a cell is changed, it can be interpreted in terms of 'signal transduction'. have. That is, it can be seen that the activated protein A 'transmits a signal' to protein B, which can be conceptually abstracted as shown on the right side of FIG. 1 (a) and expressed as a 'direction link'.

In the biochemical reaction, the inactivated form is also shown (e.g., protein B not attached to the phosphate group P), but only the activated form is mainly shown in the directed link. Numerous biochemical reactions related to signal transduction can be organized and integrated in the form of a directed link, and as a result, a complex signal transduction network structure as shown in FIG. 2 can be obtained.

In one aspect of the present invention, the basic form of signaling is a biochemical reaction in which an activated signaling molecule activates another signaling molecule. For example, protein A and protein B may be used as examples of the aforementioned signaling molecules, respectively. At this time, as shown on the left side of FIG. 1 (a), the activated protein A activates the protein B by catalyzing a reaction in which a phosphate group binds to the protein B using ATP. Therefore, usually, protein B activity can be measured as the concentration of activated protein B (e.g., phosphorylated form of protein B). The activated protein catalyzes the activation of other proteins or enters the cell nucleus to activate the expression of genes important for cell physiology.

The directed code link is a link representing a relationship between two nodes in the signaling network. Information on a direction and a sign may be mapped to one directed code link. A source of the first directed link is a departure node of the first directed link, and a target of the first directed link is an arrival node of the first directed link. That is, the direction of the directed code link is determined from the source to the target.

As shown on the right side of Fig. 1 (a), the code of the directed code link is determined according to the method of the source controlling the target, and there are mainly two methods in the signaling network. The first mode is an activation (+) mode, and the second mode is an inactivation (-) mode.

With respect to the directional code link, for example, a catalytic protein that catalyzes an actual biochemical reaction is referred to as a 'source' of the directional code link, and a protein to be catalyzed is referred to as a 'target' of the directional code link. can call

Referring to the right side of FIG. 1A, for each indicated coded link, Node A is a source node of the directional coded link and Node B is a target node of the directional coded link. In this specification, the source node may be referred to as a source node, and the target node may also be referred to as a target node.

The source node is a node corresponding to the starting point of the directed coded link. In the signaling network, the source node increases or decreases the activity of the target node according to the directed code link control method.

The target node is a node corresponding to the destination point of the directed coded link. In the signaling network, the activity of the target node is changed by the source node.

The signaling molecule may exist in an activated state (active state), and when activated, it may perform various functions that cannot be observed before activation. For example, activated signaling molecules can activate other types of signaling molecules, or they can promote gene expression to bring about changes in cell physiology.

Accordingly, the activity is a concept representing the 'extent' or 'quantity' of the signaling molecule in the active state, and can be defined in various ways. For example, activity can be defined as an intracellular concentration (eg, molarity (M)) of a signaling molecule in an active state, or as an abstract variable (a variable with arbitrary unit) without a specific physical unit. can

For example, when the activity is defined as an intracellular concentration, the activity may have a value of 0 or more.

For example, when the activity is defined as a log value of the intracellular concentration, the activity may have a real value including a negative number. In this case, in particular, the activity may be referred to as 'log activity', but may simply be referred to as 'activity'.

The term 'signal' is a concept that can be interpreted differently depending on the biological system to be analyzed. However, from the viewpoint of biological signal transduction, it can be mainly interpreted or regarded as “concentration of signal transduction molecules capable of signal transduction” as 'signal strength' or 'signal strength'.

According to one aspect of the present invention, when the biochemical reaction equation of signal transduction is abstracted with the directed link, the more abstract concept of 'activity' instead of directly corresponding the concept of signal to the concentration of signaling molecules can be dealt with as That is, it can be considered that the activity (=signal) of one node (=source node) affects the activity (=signal) of another node (=target node) through the directed code link, and in this way, one signal A phenomenon that affects a signal can be expressed as 'a signal flows'. In other words, it can be said that a signal flows from the source node to the target node.

According to the present invention, signals that are distinguished from each other may be mapped to each node. That is, the first signal may be defined by being mapped to the first node, and the first signal may be a value representing the first activity, which is the activity level of the first node. In addition, a second signal may be mapped to the second node and defined, and the second signal may be a value representing a second activity, which is the activity level of the second node.

That is, a plurality of 'signals' may be defined in the signaling network covered by the present invention. In addition, each signal is defined differently for each node.

Therefore, in the present invention, the first signal corresponding to the first node may be defined as the activity of the first node.

As described above, activity in the signal transduction network refers to the degree to which a specific signaling molecule is activated to transmit a signal. The activity of a specific molecule affects the change in the activity of other molecules, which can be expressed as 'flowing' in order to increase or decrease the signal of another molecule.

The concept of 'a signal flows' presented in the present invention may be interpreted as a concept that 'a change in the magnitude of the first signal causes a change in the magnitude of the second signal'. In addition, the meaning of "a signal flows from the source node to the target node" as described above means "a change in the signal corresponding to the source node (= activity of the source node) affects the change in the signal of the target node (= activity of the target node)" It can also mean "to give".

In the present invention, 'signal flow' may refer to a phenomenon or an effect that the activity of one signaling molecule has on the activity of another signaling molecule. The signal flow may be defined as shown in (b) of FIG. 1 or may be defined as in Equation 5 to be described later.

That is, F _BA shown in FIG. 1B is a symbol indicating the signal flow from the node A to the node B. In the present invention, _{the magnitude of the signal flow (F BA} ) is proportional to the magnitude of the activity of the node A, which is a source node, and is mapped _{to a directed code link (L BA ) having node A as a source node and node B as a target node.} It can be defined as being proportional to the weight W _{BA .} The weight W _BA may be regarded as being mapped to the directed coded link L _BA or may be regarded as being mapped to the signal flow F _BA . Hereinafter, a weight mapped to the directional coded link may be referred to as a directional coded link weight.

Thus, the signal flow (F _BA) is defined as the product of a directed link code weight (W _BA) and the activity log _(A x) of the source (F _{BA BA} = W × x _A). As shown in (c) of FIG. 1 , there may be four types of such a signal flow according to the sign combination of the link weight and the log activity of the source. In this case, the code of the signal flow may be different from that of the directed code link. For example, the log activity of the source has a negative value, whereas if the directed code link has a positive value (ie, a link that activates the target), the signal flow has a negative value.

As described above, that the log activity of the source has a negative value means that, for example, the log value of the intracellular concentration of the signaling molecule in the active state representing the source node or the log value of the value estimated from the concentration is It may mean that it has a negative value.

The present invention relates to a method for estimating the values of the weights in a situation in which structural information of the signaling network is given but specific values of the weights are not given.

As shown in FIG. 2 , the signaling network can be modeled using a model having a plurality of nodes and a plurality of directed code links as elements. The value of each element of the signal transduction network may be determined using biological experimental data. That is, it is possible to define a parameter indicating a value of each element of the signal transmission network, and fit the parameter value through a training process using the experimental data. If the experimental data for the training is sufficiently provided, the activity (=signal) of the signaling molecule of interest (=node) can be determined through the training.

However, although the topology composed of the plurality of nodes and the plurality of directed link links has been studied a lot, there is a problem in that data for the training according to the biological research result is insufficient.

Although not directly related to the present invention, related information that can help understand the present invention is provided in Korean Patent Application No. 10-2012-0098296, Attractor Landscape Analysis Reveals Feedback Loops in the p53 Network That Control the Cellular Response to DNA Damage (www .SCIENCESIGNALING.org 20 November 2012 Vol 5 Issue 251 ra83), and Control and Coordination in Biochemical Networks (IEEE Control Systems Magazine, August 2004).

The contents described so far are provided as background information to help the understanding of the present invention, and all of the above contents cannot be recognized as being disclosed prior to the time of the present application.

An object of the present invention is to provide a method for determining a control target node to be controlled in order to control an output node of a signal transmission network. Specifically, an object of the present invention is to provide a method for determining the control target node using influence, which is defined as the effect that a change in the activity of an arbitrary first node has on a change in the activity of another second node. Here, the control target node means a node that is a drug target.

In the present invention, in order to determine the activity (=signal) of a signaling molecule of interest (=node) belonging to the signaling network even in a situation in which biological experimental data is lacking, the weight of each directed code link of the signaling network is determined. We want to provide an estimation technique.

When the weight is determined, the signal flow may also be determined. Therefore, in the present invention, it is an object to provide a technique for estimating the signal flow even in a situation in which biological experimental data is lacking.

In the algorithm provided according to an aspect of the present invention, it is assumed that the activity (or signal) of a specific node is determined by the product of the activity as shown in Equation 1.

[Formula 1]

Here, a _i (t) ∈ R denotes the activity of node i at time t, a _b (i) ∈ R denotes the basal activity (or base signal) of node i, and α ∈ R is a hyperparameter. By taking the logarithm of both sides of Equation 1, it can be expressed more simply as a linear difference equation as shown in Equation 2.

[Formula 2]

Where x is the log (a) shows a ∈R ^N, b is log (a _b) shows a ∈R ^{N, N ㅧ} W∈R and ^N represents the weight matrix (weight matrix). Assuming that the signal transmission network is in a stable state over time, that is, a steady-state in which there is no change in each signal over time, the value of x in a steady state can be obtained from Equation 2 . The value of x in a steady state can be calculated by Equation 3 below.

[Equation 3]

Here, x _s ∈R ^N represents the value of x in a steady state.

The algorithm provided according to an aspect of the present invention aims to estimate the signal flow by calculating the weight of the directed code link using only the structural information of the signaling network, and predict the direction of change in the activity of major signaling molecules. . Therefore, estimating the weight using only the structure information of the network is one of the important processes, which is shown in Equation 4.

[Equation 4]

Here, A∈R ^N ㅧN is an adjacency matrix, and D _in and D _out are diagonal matrices, respectively, the number of input links (in-degree; in-degree) and number of output links (out-degree) of the i-th node. has a value as an element. The number of input links of the i-th node can be obtained by adding all the values of the i-th row of the adjacent matrix A (D _in ), and the number of output links of the i-th node can be obtained by adding all the values of the i-th column of the adjacent matrix A can (D _out ).

The algorithm provided according to one aspect of the present invention estimates the signal flow F changed by the input signal (ie, input signal) transmitted through the cell membrane receptor or the perturbation of a specific signal molecule (mainly inhibition by a drug), and the steady state It aims to predict the activity of biomolecules in F _{ij, which} is the (i, j) element of the matrix F representing the signal flow, can be expressed as the product of the link weight and log activity, as shown in FIG. 1 (b) and Equation 5.

[Equation 5]

The matrix F representing the signal flow may change with time t. This is because, as in Equation 5, F is dependent on x(t). In this case, it is assumed that W does not change with time.

However, in the steady state, x does not change with time, and therefore F does not change with time. F can be positive or negative because x is the logarithm of activity a. That is, when the activity of the signal source node decreases in the directed code link having a negative sign, as a result, the signal flow F _ij has a positive value. As a result, F having a positive value plays a role in increasing the activity of the target signaling molecule. The sign of the signal flow determined by the weight of the directed code link and the activity of the source node may be different from the sign of the directed code link.

According to one aspect of the present invention, in a signaling network defined including a plurality of nodes and a plurality of links, a method for calculating a weight mapped to each link may be provided. The method includes the steps of: obtaining, by a computing device, structural information of the signaling network, including information about a source node and a target node connected to each link, and information about a sign of a weight mapped to each link ; and a weight generation step of generating, by the computing device, a value of a weight mapped to each link by using the structure information.

In this case, the weight generating step includes, for each link, a value of a weight mapped to the link, a first number that is the number of a first set of links using a source node of the link as a source, and a target of the link determining the node as a representative value of a second number that is the number of links in a second set of targets.

In this case, the weight generation step includes, for each link, a value of a weight mapped to the link, a first number that is the number of target nodes in a first set using a source node of the link as a source, and The method may include determining the target node as a representative value of a second number that is the number of source nodes of a second set as a target.

In this case, the representative value may be a reciprocal of a geometric mean of the first number and the second number.

In this case, in the weight generation step, for each link, a value of a weight mapped to the link is determined as a value that is an inverse of a first number that is the number of a first set of links using a source node of the link as a source. or, determining the target node of the link as a value that is an inverse number of a second number that is the number of links in the second set as a target.

In this case, in the weight generation step, for each link, the value of the weight mapped to the link is set to a value that is the reciprocal of a first number that is the number of target nodes in a first set using the source node of the link as a source. or determining the target node of the link as a value that is the reciprocal of a second number that is the number of source nodes of a second set as a target.

In this case, the weight generating step may include: generating an adjacency matrix using the structure information; generating a diagonal matrix from the adjacency matrix; and calculating a weight value mapped to each link using a weight matrix generated by performing a matrix multiplication operation on the adjacent matrix and the diagonal matrix.

In this case, the method may further include an activity calculation step of calculating a value related to the activity of each node by using the generated value of each weight.

In this case, the method may further include determining a signal flow of the signaling network by using the calculated weight for each of the links and the calculated activity for each of the nodes.

According to another aspect of the present invention, in a signal delivery network defined including a plurality of nodes and a plurality of links, a program executing a method of calculating a weight mapped to each link can be read by a computer in which is recorded. A non-transitory recording medium may be provided. In this case, the program causes the computing device to obtain structural information of the signaling network, including information on the source node and the target node connected to each link, and information on the sign of the weight mapped to each link. obtaining; and a weight generation step of generating a weight value mapped to each link by using the structure information.

According to another aspect of the present invention, in a signaling network defined including a plurality of nodes and a plurality of links, a method of calculating a weight mapped to each link is performed, comprising a processing unit A computing device may be provided. At this time, the processing unit obtains structural information of the signaling network, including information about a source node and a target node connected to each link, and information about a sign of a weight mapped to each link, and the A value of a weight mapped to each link is generated using the structure information.

In this case, the computing device further includes a storage unit or a communication unit, and the structure information of the signal transmission network is obtained from information stored in the storage unit, or from an external device of the computing device through the communication unit. may have been obtained.

According to another aspect of the present invention, in a signal transduction network defined including a plurality of nodes and a plurality of links representing a protein, a method for calculating protein activity for calculating information about the activity of a specific protein may be provided. . The method includes the steps of: obtaining, by a computing device, structural information of the signaling network, including information about a source node and a target node connected to each link, and information about a sign of a weight mapped to each link ; a weight generation step of generating, by the computing device, a value of a weight mapped to each link by using the structure information; And the computing device, using the generated value of each weight, according to the change in the activity of the first protein indicated by the first node, calculating information about the change in the activity of the second protein indicated by the second node includes steps.

In this case, the change in the activity of the first protein is the basal activity of the first protein that changes according to the administration of the first drug targeting the first protein, and the change in the activity of the second protein indicated by the second node The information about may represent the rate of change of the activity of the first protein.

According to one aspect of the present invention, it is possible to provide a method of determining a control target node to be controlled for controlling an output node of a signaling network. This determination method calculates the influence on a first output node of each of a plurality of nodes included in the first signal delivery network, and calculates an influence of a plurality of influences corresponding to each of the plurality of nodes. Luance calculation step; and a control target for controlling the first output node by selecting some of the plurality of nodes according to a control target preset for the first output node based on the values of the plurality of influences. and a control target node determining step of determining as a node. In addition, the influence of any first node among the plurality of nodes on the first output node indicates the degree of influence of the change in the activity of the first node on the change in the activity of the first output node.

In this case, the determining method includes calculating a Shortest Path Length to Output (SPLO) from each of the plurality of nodes to the first output node between the influence calculation step and the control target node determination step, The method may further include calculating a plurality of SPLOs corresponding to each of the plurality of nodes. And in this case, the step of determining the control target node includes: classifying the plurality of nodes into a plurality of SPLO groups based on the value of the SPLO; and independently for each of the plurality of SPLO groups, a group-by-group determination step of selecting at least some of the nodes belonging to each SPLO group and determining as a control target node for controlling the first output node; can and the SPLO from the arbitrary node to the first output node is one or more links connecting the arbitrary node and the first output node to each other to reach the first output node from the arbitrary node. Among one or more candidate paths that can be selected from among the one or more candidate paths, the number of links included in the shortest path having the smallest number of links included in each candidate path may be indicated.

In this case, the signal transduction network is a network that models the interaction between a plurality of biomolecules, and may be defined including a plurality of directional links and a plurality of nodes connected by the links. And each node may represent a biomolecule.

In this case, the activity of any node included in the first signal transduction network may indicate the amount of activated biomolecules represented by the arbitrary node.

In this case, the control target preset for the first output node may be to increase or decrease the activity of the first output node.

In this case, the first output node may be any one of a node representing cell death, a node representing cell proliferation, and a node representing cell metastasis.

At this time, the change in the activity of the first node is the change (∂b _{j ) of the basal activity of the first node, and the influence (S ij} ) of the first node to the first output node is the following Equation 11 can be satisfied [Equation 11] S _ij = ∂x _i / ∂b _j.

According to another aspect of the present invention, a method for determining a control target node to be controlled for controlling an output node of a signaling network may be provided. The determining method includes calculating an influence for a first output node of each of a plurality of nodes included in the first signal delivery network, and calculating a plurality of influences corresponding to each of the plurality of nodes. ; calculating an SPLO from each of the plurality of nodes to the first output node, and calculating a plurality of SPLOs corresponding to each of the plurality of nodes; classifying the plurality of nodes into a plurality of SPLO groups based on the value of the SPLO; and independently for each of the plurality of SPLO groups, selecting at least some of the nodes belonging to each of the SPLO groups and determining them as a control target node for controlling the first output node. At this time, the influence of the arbitrary first node among the plurality of nodes on the arbitrary output node is the effect of the change in the activity of the first node on the change in the activity of the arbitrary output node (∂x _{i )} is the degree of

According to another aspect of the present invention, it is possible to provide a method for determining a control target node to be controlled for controlling an output node of a signaling network. The determining method includes: for each of a plurality of output nodes included in the first signaling network: calculating an influence on the output node of each of a plurality of nodes included in the first signaling network, calculating a plurality of influences corresponding to each of a plurality of nodes, calculating an SPLO from each of the plurality of nodes to the output node, and calculating a plurality of SPLOs corresponding to each of the plurality of nodes; Classifying the plurality of nodes into a plurality of SPLO groups based on the value of the SPLO, and independently selecting at least some of the nodes belonging to each SPLO group for each of the plurality of SPLO groups and outputting the output and determining a control target candidate set for each output node, configured to be determined as a control target node for control of the node. And the determining method includes a control target node determining step of determining a final control target node for controlling the first output node based on the intersection of the control target nodes determined for each of the plurality of output nodes. . At this time, the influence of the arbitrary first node among the plurality of nodes on the arbitrary output node is the effect of the change in the activity of the first node on the change in the activity of the arbitrary output node (∂x _{i )} is the degree of

At this time, the SPLO from the arbitrary node to the arbitrary output node is at least one connecting the arbitrary node and the arbitrary output node to each other in order to reach the arbitrary output node from the arbitrary node. Among one or more candidate paths that can be selected from the links, the number of links included in the shortest path having the smallest number of links included in each candidate path may be indicated.

At this time, the influence of any first node among the plurality of nodes on the first output node is in all paths of the first signaling network from the arbitrary first node to the first output node, It may be calculated using an operation of summing the products of weights assigned to links included in each path.

In this case, the weights assigned to the links included in each path include, by the computing device, information on a source node and a target node connected to each link, and information on a sign of a weight mapped to each link, obtaining structural information of the signaling network; and a weight generation step of generating, by the computing device, a value of a weight mapped to each link by using the structure information; .

In this case, the weight generating step may include: generating an adjacency matrix using the structure information; generating a diagonal matrix from the adjacency matrix; and calculating a weight value mapped to each link using a weight matrix generated by performing a matrix multiplication operation on the adjacent matrix and the diagonal matrix.

According to one aspect of the present invention, there may be provided a method for determining a biomolecule that should be a target of a drug in order to control the activity of an output biomolecule that is a specific biomolecule in a biomolecular network composed of interactions between biomolecules. . In this method, the influence of each of a plurality of biomolecules included in the biomolecule network is calculated on the output biomolecule to calculate a plurality of influences corresponding to each of the plurality of biomolecules. Luance calculation step; and a drug target biomolecule for controlling the output biomolecule by selecting some of the plurality of biomolecules according to a control target preset for the output biomolecule based on the values of the plurality of influences and a drug target biomolecule determination step of determining as a molecule. And in this case, the influence of any first biomolecule among the plurality of biomolecules on the output biomolecule is the degree to which the change in the activity of the first biomolecule affects the change in the activity of the output biomolecule. . Here, the 'each of the plurality of biomolecules' is a collective concept and may correspond to each node of the signal cutting network described above.

The biomolecule network composed of the interactions between the biomolecules can be modeled as the aforementioned signal transduction network. The output biomolecule, which is the specific biomolecule, may refer to the above-described output node. The biomolecule to be the target of the drug may refer to the aforementioned control target node.

According to the present invention, it is possible to provide a method for determining a control target node (=drug target node) to be controlled for the control of an output node of a signaling network. Specifically, it is possible to provide a method of determining the control target node by using influence defined as an effect of a change in the activity of an arbitrary first node on a change in the activity of another second node.

According to the present invention, even in a situation in which biological experimental data is lacking, the weight of each directed code link of the signaling network can be estimated, and as a result, the signal flow can be estimated.

As a result, by using the defined signaling network, a base technology capable of determining the activity (=signal) of a signaling molecule of interest (=node) belonging to the signaling network can be provided.

According to the present invention, it is possible to discover drug target candidates using only structural information of a complex signal transduction network without kinetic experimental data requiring a lot of time and money.

The present invention has an advantage in that it is possible to preferentially select a control target that can be a drug target without experimental data on nonlinear kinetics by using only the structure information of the network. Unlike a biomedical science-related researcher who selects an arbitrary drug target based on experience, knowledge, insight, etc., the present invention enables quantitative analysis of the structural information of a vast amount of signal transduction network that has been accumulated over a long period of time. Drug target candidates may be preferentially selected.

1 shows the concept of a biochemical reaction and a directed code link in the signaling network dealt with in the present invention.
Figure 2 schematically presents an example of the structure of a complex signaling network.
3 is a diagram illustrating a structure of a computing device executing a method for generating information on a signaling network provided according to an embodiment of the present invention.
4 is a flowchart illustrating a method of generating information on a signaling network provided according to an embodiment of the present invention.
5 is a diagram illustrating an example of the structure of a signaling network defined according to an embodiment of the present invention.
6 shows an example of a table including structure information provided according to an embodiment of the present invention.
FIG. 7 shows an adjacency matrix generated using the structure information provided by the table shown in FIG. 6 .
8 shows an example of a diagonal matrix generated according to an embodiment of the present invention.
9 illustrates an example of a weight matrix generated according to an embodiment of the present invention.
10 shows an example of a signaling network having a 3-node cascade structure.
11 shows an example of a signaling network having a feed-forward structure having three paths.
12 shows an example of a signaling network having a three-node negative feedback loop structure.
13 shows an example of a signaling network having a loop unrolling structure for a 3-node voice feedback loop.
14 shows an EMT control signaling network.
15 and 16 show examples of influence calculation results for the EMT control network shown in FIG. 14 .
17 is a flowchart illustrating a method of determining a control target in a signaling network according to an embodiment of the present invention.
18 is a flowchart illustrating a process of selecting a control target node according to an embodiment of the present invention.
19 is a flowchart illustrating a process of selecting a control target node according to another embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, the present invention is not limited to the embodiments described herein and may be implemented in various other forms. The terminology used in this specification is intended to help the understanding of the embodiments, and is not intended to limit the scope of the present invention. Also, singular forms used hereinafter include plural forms unless the phrases clearly indicate the opposite.

3 is a diagram illustrating a structure of a computing device executing a method for generating information on a signaling network provided according to an embodiment of the present invention.

The computing device 100 may include a processing unit 101 made of a CPU or the like, a storage unit 102 , and a communication unit 103 .

Each of the steps included in the method for generating information about the signal transmission network to be described below may be executed by the computing device 100 .

4 is a flowchart illustrating a method of generating information on a signaling network provided according to an embodiment of the present invention.

5 is a diagram illustrating an example of the structure of a signaling network defined according to an embodiment of the present invention.

Hereinafter, it will be described with reference to FIGS. 3 to 5 together.

<Calculation of weights>

The signaling network 1 may be defined to include a plurality of nodes 2 and a plurality of links 3 .

Each node 2 may represent a molecule in which an organism exists in a cell. The molecule may be the aforementioned signal transduction molecule. The number indicated in the circle representing each node 2 is the node identifier of the corresponding node.

Each link 3 may have the form of the above-described directional code link. Each link 3 may represent an interaction between the source node and the target node.

Each link 3 may be divided into a first type 31 symbolized by an arrow-shaped leader line and a second type 32 symbolized by a nail head-shaped leader line. The first type (31) is a type in which an increase in a signal at a source node activates an increase in a signal at a target node, and in the second type (32), an increase in a signal at a source node suppresses an increase in a signal at the target node. It may be of the type that

A weight may be assigned to each link. The weight may be a real number having a positive value or a negative value.

An embodiment of the present invention may aim to determine a specific value of the weight.

A method for generating information on a signaling network provided according to an embodiment of the present invention may include the following steps.

In step S10 , the processing unit 101 of the computing device 100 obtains the structure information of the signal transmission network from the storage unit 102 , or an external storage device or server through the communication unit 103 . It can be obtained from other network entities outside such as

The structure information may include information about a source node and a target node, which are two nodes 2 connected to each link 3 , and information about a sign of a weight mapped to each link 3 . . However, the structure information does not include a specific value of the mapped weight.

6 shows an example of a table including structure information provided according to an embodiment of the present invention.

A key for distinguishing different records (rows) of the table 50 may be an identifier of each link 3 .

In the table 50, a first field relating to the identifier (Ns) of the source node of each link, a second field relating to the type of each link, and a first field relating to the identifier (Nt) of the target node of each link It can contain 3 fields.

In the second field, when the link of the corresponding record is the first type (21), a value indicating this may be stored, and when the link of the corresponding record is the second type (22), a value indicating this may be stored. .

In the weight generation step S20 , the processing unit 101 of the computing device 100 may generate a weight value mapped to each link 3 using the structure information.

In the first embodiment of the present invention, in the weight generation step ( S20 ), the value of the weight mapped to the link 3 is the number of links in the first set using the source node of the link 3 as a source. and determining ( S201 ) as a representative value of the first number and the second number, which is the number of links in a second set targeting the target node of the link. The step S201 may be executed for each link 3 .

For example, in the signaling network 1 shown in FIG. 5 , the step S201 may be performed for the second links 3 and L2. In this case, the value of the weight mapped to the second link (3, L2) is the value of the first set of links (L2) whose source is the node (N2), which is the source node of the second link (3, L2). A first number (=1) that is the number, and a second number that is the number of the second set of links (L2, L5) targeting a node (N3) that is a target node of the second link (3, L2) ( =2) can be determined as a representative value. In this case, the representative value may be, for example, a geometric average of the first number and the second number.

In the second embodiment of the present invention, the weight generation step ( S20 ) includes the value of the weight mapped to the link 3 , of a first set of target nodes using the source node of the link 3 as a source. and determining ( S202 ) a representative value of a first number that is the number and a second number that is the number of source nodes of a second set targeting the target node of the link. The step S202 may be executed for each link 3 .

For example, in the signaling network 1 shown in FIG. 5 , the step S202 may be performed for the second links 3 and L2. In this case, the first set of target nodes N3 using the value of the weight mapped to the second link 3 and L2 as the source node N2 as the source node of the second link 3 and L2 as a source. A first number (=1) that is the number of , and a second number that is the number of the second set of source nodes N2 and N4 targeting the node N3 that is the target node of the second link 3 and L2 It can be determined as a representative value of the number (=2). In this case, the representative value may be, for example, a geometric average of the first number and the second number.

In the third embodiment of the present invention, in the weight generation step ( S20 ), the value of the weight mapped to the link 3 is the number of links in the first set using the source node of the link 3 as a source. determining as a value that is the reciprocal of the first number, or determining the target node of the link 3 as a value that is the reciprocal of a second number that is the number of links in a second set that targets the target node of the link 3 ( S203 ). can

For example, in the signaling network 1 shown in FIG. 5, the step S203 may be performed for the second links 3 and L2. In this case, the value of the weight mapped to the second link (3, L2) is the value of the first set of links (L2) whose source is the node (N2), which is the source node of the second link (3, L2). A second set of links (=1) that is determined as a value (=1) that is the reciprocal of the first number (=1), or that targets a node N3 that is a target node of the second link (3, L2) ( It can be determined as a value (=1/2) that is the reciprocal of the second number (=2) that is the number of L2 and L5).

In the fourth embodiment of the present invention, the weight generating step ( S20 ) includes the value of the weight mapped to the link 3 , of the first set of target nodes using the source node of the link 3 as a source. The step of determining as a value that is the reciprocal of the first number that is the number, or determining the target node of the link 3 as a value that is the reciprocal of the second number that is the number of the second set of source nodes targeting the target node (S204) may include

For example, in the signaling network 1 shown in FIG. 5 , the step S204 may be performed for the second links 3 and L2. In this case, the first set of target nodes N3 using the value of the weight mapped to the second link 3 and L2 as the source node N2 as the source node of the second link 3 and L2 as a source. The second set of source nodes is determined as a value (=1) that is the inverse of the first number (=1) that is the number of , or targets the node N3, which is the target node of the second link 3, L2. It can be determined as a value (=1/2) that is the reciprocal of the second number (=2) that is the number of the items N2 and N4.

In the fifth embodiment of the present invention, the weight generating step (S20) includes: generating an adjacency matrix (A) using the structure information (S211); generating a diagonal matrix (D) from the adjacent matrix (A) (S212); and calculating the weight value mapped to each link using the weight matrix W generated by performing matrix multiplication operation on the adjacency matrix A and the diagonal matrix D according to the weight generation method (S201 to S204). It may include a step (S213) of doing.

A specific method of generating the adjacency matrix A of the fifth embodiment ( S211 ) will be described with reference to FIG. 7 .

FIG. 7 shows an adjacency matrix A generated using the structure information provided by the table 50 shown in FIG. 6 .

The number of rows and the number of columns of the adjacent matrix A are the same as the number of nodes included in the signal transmission network 1, respectively. 5 to 7, since the number of nodes included in the signaling network 1 is 4, the adjacency matrix A becomes a 4x4 square matrix.

All elements of the adjacency matrix A may be initialized to a value of '0'.

Then, among the elements included in the adjacency matrix A, the values of N elements equal to the number N of links included in the signaling network 1 are replaced with values of +1 or -1. .

In this case, the position (s, t) of the element replaced with a value of +1 or -1 is determined by the identifier index (s) of the source node of each link and the identifier index (t) of the target node. And the substituted value is determined by the type of each link.

For example, in the table 50 of FIG. 6 , the identifier index s of the source node of the first link L1 whose link ID is L1 is 1, and the identifier index t of the target node of the first link L1 is ) is 2. Accordingly, in the adjacent matrix A shown in FIG. 7 , an element at a position where the s-th column (= first column) and the t-th row (= second row) intersect is substituted with +1 or -1. In this case, the substituted value is substituted with +1 because the link type of the first link L1 is the first type.

As another example, in the table 50 of FIG. 6 , the identifier index s of the source node of the third link L3 whose link ID is L3 is 3, and the identifier of the target node of the third link L3 is The index (t) is 1. Accordingly, in the adjacent matrix A shown in FIG. 7, an element at a position where the s-th column (= third column) and the t-th row (= first row) intersect is substituted with +1 or -1. In this case, the substituted value is substituted with -1 because the link type of the third link L3 is the second type.

Hereinafter, an example of the step S212 of generating the diagonal matrix D of the fifth embodiment will be described. The fact that the diagonal matrix D generated by the adjacency matrix A shown in FIG. 7 is given as shown in FIG. 8 can be easily understood from the definition of the diagonal matrix described below in Equation 4 above. The diagonal matrix D may include an input diagonal matrix Din and an output diagonal matrix Dout.

8 shows an example of a diagonal matrix generated according to an embodiment of the present invention.

Hereinafter, an example of the step S213 of calculating the weight value of the fifth embodiment will be described. By calculating the matrix product of the adjacency matrix A shown in FIG. 7 and the diagonal matrix D shown in FIG. 8 so that the representative value is the inverse of the geometric mean of the first number and the second number in the weight generation step S201 (refer to Equation 4), the weight matrix W illustrated in FIG. 9 can be obtained. In this case, for example, the W ₂₁ part of the weight matrix W represents the weight value of the first link L1.

9 illustrates an example of a weight matrix generated according to an embodiment of the present invention.

In one embodiment of the present invention, when using a process of generating weights by dividing by degrees of nodes connected to a link (or by dividing by geometric mean values of Indigri and Outdegree) to generate a weight matrix, the above process is This can be called link weight normalization. Although the process of generating the weight itself is not a process of normalizing the link weight, the weight can be generated through the process of normalizing the link weight.

The weight matrix generated by executing step S223 may be the same as the weight matrix generated by executing step S213.

<Influence>

In an embodiment of the present invention, the weight matrix may be further utilized.

In the present invention, the term 'influence' refers to the degree of influence of a change in the activity of any other node on the activity of the output node through the signal flow when a specific output node is designated in the signal delivery network. , and the _{measure of the influence (S ij} ) is defined as in Equation 6.

[Equation 6]

S _ij = ∂x _i / ∂b _j = (∂x _i / ∂x _j )·(∂x _j / ∂b _j )

Here, x _i denotes the activity of the i-th node having the index i, and b _j denotes the basal activity of the j-th node j having the index j. The i-th node may be referred to as an output node.

Therefore, the influence (S _ij ) is a value representing the rate of change in the activity of the i-th node according to the change in the basal activity of the j-th node (j).

Mathematically, the influence (S _ij ) means the rate of change or partial derivative of _{the activity (x i} ) of the i-th node (i) with respect to the change in _{the basal activity b j} of the j-th node (j). . For example, influence is a protein that regulates cell death or proliferation when the basal activity of a certain signaling molecule (j-node (j)) is changed due to mutation or drug (output node, i-node (i)) It indicates how much the activity of (ie, rate of change) changes.

The value of influence can be interpreted in terms of sign and magnitude. _{If the influence (S ij} ) of a certain j-th node (j) on the output node i-th node (i) is negative, if the activity of the j-th node (j) increases, the output node i-th node (i) ) means that the activity decreases. Conversely, when the influence (S _ij ) is a positive number, it means that when the activity of the j-th node (j) increases, the activity of the i-th node (i), which is an output node, increases. For example, if it is desired to reduce the activity of a protein related to cell proliferation by inhibiting a certain signaling protein to decrease its activity, a signaling protein having a positive influence on the cell proliferation related protein should be selected. And since it means that the larger the size of the influence, the greater the effect on the output node. Therefore, the sign of the influence is determined according to the goal to control the cell state, and the priority of the control target nodes according to the size of the influence. (priority) or rank (rank) can be assigned.

When a drug targeting the j-th node j is administered, the basal activity b _j of the j-th node j, which is the drug target node, will change. So, one can observe how the activity of the j-th node (j) changes in the basal activity (b _j), and the result of the i-node (i) the output node as the administration of the drug varies.

<Method of calculating influence using weights 1>

In the present invention, the concept of influence defined above is applied to estimate the activity of each node. First, use Equation 7 below.

[Equation 7]

where ^{x∈R N} , W∈R ^N×N , α ∈R .

The difference between Equation 7 and Equation 2 is that β ( β ∈R) is introduced instead of the (1-α ) term, which is Wx(t), which represents the effect of signal flow in Equation 7 to make the existing equation more general. The constraint that the sum of weights α + β used in the sum of weights of b and b indicating the effect of basal activity is always 1 is relaxed. (On the other hand, in Equation 2, when the value of α is determined, β is automatically determined as 1- α. do). However, in general , setting α and β so that α + β = 1 guarantees the stability of the numerical analysis that solves the algorithm.

For x(t) in Equation 7, the influence matrix S is derived as Equation 8.

[Equation 8]

In this embodiment, the influence S(∞) in the steady-state is used. In fact, in the numerical analysis process to obtain S(∞), the values of the matrix norm, ||S(t+1) - S(t)||, converge below a ^{specific value (eg, tol=1e -6 ).} Use the value of S(t) as an approximation of S(∞).

Hereinafter, a method of calculating the influence will be described with reference to FIG. 10 . 10 is for explaining the concept of calculating the value of influence.

10 shows an example of a signaling network having a 3-node cascade structure.

10 is one of the simplest structures as a signaling cascade in which three nodes are sequentially connected. The influence (S ₂₁ ) of the node x _{1 on} the node x ₂ is calculated as in Equation 9. In Equation 9, b ₁ represents the basal activity of the node x _{1 and b 2} represents the basal activity of the node x ₂ .

[Equation 9]

Going one step further, the influence (S _{31 ) of the node x 1 on} the node x ₃ is as in Equation 10.

[Equation 10]

Hereinafter, a method of calculating the influence will be described with reference to FIG. 11 . 11 is for explaining the concept of calculating the value of influence.

11 shows an example of a signaling network having a feed-forward structure having three paths.

11 is _{a feedforward structure having three paths from a node x 1} to x ₆ , and the influence calculation therefor is as shown in Equation 11.

[Equation 11]

What we can know from the influence calculation formulas for the structures of FIGS. 10 and 11 is that the influence (S _ij ) is the product of the link weights on each path in all paths from the source node j to the output node i ( It is the result of summing ΠW). For example, S ₆₁ of FIG. 11 has three terms made up of a product of a weight W, and it can be seen that each term corresponds to each path on the feedforward of FIG. 11 . That is, for example, W ₆₄ ·W ₄₂ ·W ₂₁ corresponds to the path x ₁ - x ₂ - x ₄ - x _{6 .}

Hereinafter, a method of calculating the influence will be described with reference to FIGS. 12 and 13 . 12 and 13 are for explaining the concept of calculating an influence value.

12 shows an example of a signaling network having a three-node negative feedback loop structure.

And FIG. 13 shows an example of a signaling network having a loop unrolling structure for a 3-node voice feedback loop. In FIG. 13 , t determines the degree of unrolling.

The influence calculation for the voice feedback loop structure of FIG. 12 is slightly more complicated than the cascade or feedforward structure. Since such a structure includes a loop or a cycle, the path can be considered infinitely. When node redundancy is allowed in FIG. 12 , paths from node 1 to node 3 become infinitely numerous due to the cycle. In a steady state, the influence is obtained by assuming t → ∞, which can be considered as calculating the influence for a path of infinite length from the perspective of loop unrolling (refer to FIG. 13). However, in actual numerical analysis for influence calculation, the value of ||S(t+1) - S(t)|| becomes very small, so the number of iterations of the loop for influence calculation can be limited to a certain natural number T. have. In some cases, the influence calculation diverges depending on the weight W and the values of the parameters α and β . In this case, the number of repetitions of the loop for the influence calculation may be designated as T _{L, which is a natural number of a specific value.} Empirically, W is obtained by the above-described weight estimation algorithm, and α = 0.9 and β = 0.1 may be set.

<Method of calculating influence using weights 2>

In the first embodiment of obtaining the influence (S _{ij ), the influence (S ij} ) may be given as in Equation (12).

[Equation 12]

S = (1- α )*( I + αW + α ² W ² + ... + α ^{t -1} W ^{t -1} )

Equation 12 shows the case where β = 1- α in Equation 8.

In Equation 12, α is a hyperparameter that the user can specify and input. And W denotes the above-described weight matrix. Accordingly, the matrix of the influence values (S∈R ^N×N ) may be calculated using the weights. As a result, when a drug targeting the j-th node (j) is administered, it can be estimated how the activity of the i-th node, which is the output node, changes.

<Calculation of activity using influence>

In step S30 , the value of the above-described influence S _ij may be calculated _{using the above-described weight W ij and Equation 7 above.} In addition, a value related to the activity (x _i ) of each node (i) may be calculated using the calculated value of the influence (S _{ij ).}

In step S30, Equation 2 or Equation 7 may be used to calculate a value related to the activity of each node.

In step S40, using the weight (W _{ij ) calculated for each of the links (3) and the activity (x i} ) calculated for each of the nodes (2), signal flow can be determined.

<Excavation of control targets using SI-plot>

By analyzing the shortest path length to output (hereinafter, SPLO) of the output node together with the influence described above, the present inventors can discover effective control targets that can control the proliferation or death of cells. found out that

14 shows an EMT control signaling network.

The EMT control signaling network has an EMT node as its only output.

15 and 16 show examples of influence calculation results for the EMT control network shown in FIG. 14 .

15 is an influence calculation result for the EMT control signaling network in FIG. 14 .

In FIG. 15 , the horizontal axis indicates an influence value for an EMT node among the nodes of the EMT control network, and the vertical axis indicates an identifier of each node of the EMT control signaling network. For example, the influence on the EMT node of the TWIST1 node has a value between 1 and 2, the influence on the EMT node of the EGFR node has a value between 0 and 1, and the influence on the EMT node of the E-Cadhorin node. It can be seen that s has a value of about -3. The influence value may be a real number.

In FIG. 15, the influence calculation results are arranged in descending order based on the influence value. The number next to the bar graph shown in FIG. 15 indicates the shortest path length (SPLO) for the EMT output node. The shortest path may be a natural number. For reference, the Pearson correlation coefficient between influence and SPLO is -0.753.

In the present invention, the Shortest Path Length (SPL) from the first node to the second node is one connecting the first node and the second node to each other in order to reach the second node from the first node. Among one or more paths that can be selected from the above links, the number of links included in the shortest path having the smallest number of links included in each path is indicated.

16 shows influence together with SPLO, which may be referred to as an SI-plot (SPLO-Influence plot) below.

The SI-plot has a horizontal axis and a vertical axis. The horizontal axis represents the SPLO value from the given node to the EMT node, and the vertical axis represents the influence value of the given node to the EMT node. The horizontal axis may represent a natural number, and the vertical axis may represent a real number. The value of SPLO was set to increase from the horizontal axis to the right, and the value of influence was set to increase from the vertical axis to the top.

If the first SI-plot relates to a first output node, the second SI-plot different from the first SI-plot may relate to a second output node different from the first output node. For example, FIG. 16 is an SI-plot for an EMT node.

In the SI-plot for the EMT node shown in FIG. 16 , information indicating other specific nodes other than the EMT node may be displayed. For example, the TWIST1 node may be displayed in the SI-plot of FIG. 16 , and the displayed position is a position corresponding to the SPLO from the TWIST1 node to the EMT node and the influence value (S _{EMT_TWIST1} ) of the TWIST1 node for the EMT node.

From the above description, it can be understood that nodes other than the first output node may be plotted in the first SI-plot for the first output node. In addition, the vertical axis of the first SI-plot represents the influence value of the first output node for a given arbitrary node, and the horizontal axis represents the SPLO from the given arbitrary node to the first output node.

In FIG. 16 , the node indicated by adding an underscore to the name of the node (the node indicated by a triangle) is the result of single inhibition or dual inhibition by the present inventors for a Boolean model built based on detailed logic information. It is a 'control target node' found by searching under all conditions.

It can be easily seen from FIG. 16 that when the absolute value of influence is large, the SPLO tends to be small (Pearson's correlation coefficient -0.753). This means that the closer to the output node (the EMT node is the output node in FIG. 16), the greater the influence, and it can be seen as a result that is easy to understand by common sense.

However, there are nodes that do not follow this trend. For example, ZEB2 has SPLO=2 but has relatively less influence than nodes with the same SPLO. Conversely, in the case of NOTCH (SPLO = 5) or RAS (SPLO = 6), the influence is relatively larger than that of nodes having the same SPLO.

It will be described below with reference to FIG. 16 .

In order to analyze the case where the closer to the output node (EMT node) the influence is not necessarily large, the present inventors displayed the influence together with SPLO as shown in FIG. 16 (SI-plot). As a result, it was found that the 'control target nodes' discovered by the present inventors through Boolean simulation analysis and biological experiment verification were located at the top of each SPLO value in the SI-plot. In addition, according to the embodiment, it was found that the discovered 'control target nodes' tend to be distributed in a diagonal direction from the upper left to the lower right. This means that even if the absolute value of influence is not large, if the influence is large within the same SPLO group, it is highly likely to become a control target node.

This means that the absolute value of influence is important from a control point of view, but the relative size of influence within each SPLO group is also important. The present inventors analyzed this trend in the blood cancer cell-related signaling network (Zanudo et al. PLoS Comp. Biol. 11(4), e1004193) and the colorectal cancer signaling network of various scales (Fumia et al. PLoS ONE, 8 ( 7), e69008, Cho et al., BMC Sys. Biol., 10(1), 96).

The method of discovering a control target in the present invention based on the results discussed above is a method of selecting nodes with high influence in each SPLO group as a control target after creating an SI-plot by calculating the influence and SPLO. The method of excavating the control target in the present invention is summarized as the flowchart shown in FIG. 17 .

<Method of determining signal flow control target for single output node>

17 is a flowchart illustrating a method of determining a control target in a signaling network according to an embodiment of the present invention.

In step S110 , the computing device receives the first signaling network and the first output node of the first signaling network as inputs.

In step S120, the computing device calculates the influence on the first output node of any node included in the first signal delivery network, and one or more or a plurality of the first output nodes Calculate the influence value of dogs.

In step S130, the computing device calculates a plurality of SPLOs for the first output node from each of the arbitrary nodes.

In step S140 , the computing device generates a first SI-plot for the first output node based on the calculated plurality of influence values and the plurality of SPLOs. In this case, generating the first SI-plot may mean generating data necessary for generating the first SI-plot.

It can be understood that the first SI-plot can be generated using the above-described steps S110 to S140 with respect to the first output node. Also, it will be understood that a second SI-plot different from the first SI-plot can be generated by applying the steps S110 to S140 for a second output node different from the first output node can

In step S150, a node is selected according to a control target for each SPLO group in the first SI-plot. The selected nodes may be referred to as 'control target nodes'.

In this case, each of the SPLO groups may be distinguished by the value of the horizontal axis of the first SI-plot. That is, nodes having the same SPLO as each other among the nodes plotted in the first SI-plot may be regarded as belonging to the same SPLO group. In addition, if the SPLO of the first node and the SPLO of the second node plotted in the first SI-plot are different from each other, the first node and the second node may be regarded as belonging to different SPLO groups.

Here, the criterion for selecting the control target node may be independently set for each SPLO group.

For example, if the control goal is to lower the activity of the first output node by lowering the activity of a specific node plotted on the first SI-plot, the sign of the influence in each SPLO group is positive (+). , it is possible to select a node having an absolute value of the influence greater than a preset threshold and determine it as the control target node. Quantitatively, "a node having an absolute value of influence greater than a preset threshold" may be defined as a node having an influence greater than a user-defined threshold of influence. For example, it can be defined as “top 3 within each SPLO group” or “top 5% within each SPLO group”.

As another example, if the control goal is to increase the activity of the first output node by lowering the activity of a specific node plotted on the first SI-plot, the 'control target node' may be selected based on a different criterion than the above. can

<Method of determining signal flow control target for multiple output nodes>

The control of a signaling network with multiple output nodes is not much different from the control of a network with only one output node. After obtaining an SI-plot for each output node, a control target is selected according to the control target.

For example, a signaling network may have three output nodes, such as a first output node indicating apoptosis, a second output node indicating proliferation, and a third output node indicating metastasis. In this case, the goal of normal control is to increase the activity of the death output node and decrease the activity of the proliferative output node and the transition output node through drug inhibition.

In this case, a node that has negative influence on the death output node and positive influence on the proliferation output node and transition output node should be determined as the 'control target node'.

18 is a flowchart illustrating a process of selecting a control target node according to an embodiment of the present invention.

In step S210, a first SI-plot that is an SI-plot for a dead output node, a second SI-plot that is an SI-plot for a proliferating output node, and a third SI-plot that is an SI-plot for a transitional output node information to create

In step S220, 'control target nodes' are selected from the first SI-plot to generate a first control target candidate set including the selected 'control target nodes'. At this time, the influence of each of the 'control target nodes' has a negative sign and its magnitude (absolute value) belongs to a higher rank in each SPLO group of the first SI-plot.

In step S230, 'control target nodes' are selected from the second SI-plot to generate a second control target candidate set including the selected 'control target nodes'. At this time, the influence of each of the 'control target nodes' has a positive sign, and its magnitude (absolute value) belongs to a higher rank in each SPLO group of the second SI-plot.

In step S240, 'control target nodes' are selected from the third SI-plot to generate a third control target candidate set including the selected 'control target nodes'. At this time, the influence of each of the 'control target nodes' has a positive sign, and its magnitude (absolute value) belongs to a higher rank in each SPLO group of the third SI-plot.

In steps S220 , S230 , and S240 , “belonging to the upper rank” may mean that the size of the corresponding influence is greater than or equal to a predetermined threshold.

In step S250, one or more 'final control target nodes' may be selected from the intersection of the first control target candidate set, the second control target candidate set, and the third control target candidate set. The 'final control target node' may finally become a drug target.

If the process of the steps S210 to S250 is generalized, it can be expressed as in the flowchart shown in FIG. 19 .

19 is a flowchart illustrating a process of selecting a control target node according to another embodiment of the present invention.

In step S310, an SI-plot is generated for each of a plurality of output nodes defined in a specific signaling network to generate a plurality of SI-plots.

In step S320, for each SI-plot, the signal flow control target determination method for a single output node is applied to each output node according to the control target set for each output node of each SI-plot. Obtain a control target candidate set. When there are N output nodes, N number of control target candidate sets may also be generated.

In step S330, the intersection of the control target candidate sets may be obtained, and one or more of the nodes included in the intersection may be selected to determine the final control target node.

By using the above-described embodiments of the present invention, those skilled in the art will be able to easily implement various changes and modifications within the scope without departing from the essential characteristics of the present invention. The content of each claim in the claims may be combined with other claims without reference within the scope that can be understood through this specification.

Claims (14)

A method for determining a control target node to be controlled for controlling an output node of a signal transmission network, the method comprising:
The computing device calculates the influence on the first output node of each of the plurality of nodes included in the first signal transmission network, and calculates the influence of the plurality of influences corresponding to each of the plurality of nodes. word calculation step; and
The computing device controls the first output node by selecting some of the plurality of nodes according to a control target preset for the first output node based on the values of the plurality of influences A control target node determining step of determining as a control target node for;
includes,
The influence of any first node among the plurality of nodes on the first output node is a degree of an effect that a change in the activity of the first node has on a change in the activity of the first output node,
The change in the activity of the first node is the change in the basal activity of the first node (∂b _j ),
The influence of the first node on the first output node (S _ij ) satisfies Equation 1 ([Equation 1] S _ij = ∂x _i / ∂b _j ),
Determination method of control target node. According to claim 1,
The computing device calculates a Shortest Path Length to Output (SPLO) from each of the plurality of nodes to the first output node between the step of calculating the influence and the step of determining the control target node, and the plurality of Further comprising the step of calculating a plurality of SPLOs corresponding to each of the nodes,
The control target node determination step is,
classifying, by the computing device, the plurality of nodes into a plurality of SPLO groups based on the value of the SPLO; and
A group-by-group determination step in which the computing device independently selects at least some of the nodes belonging to each SPLO group for each of the plurality of SPLO groups and determines as a control target node for controlling the first output node ;
includes,
The SPLO from any of the nodes to the first output node includes one or more links connecting the arbitrary node and the first output node to each other to reach the first output node from the arbitrary node. represents the number of links included in the shortest path with the smallest number of links included in each candidate path among one or more candidate paths that can be selected from
Determination method of control target node. According to claim 1,
The signal transduction network is a network that models interactions between a plurality of biomolecules, and is defined by including a plurality of directional links and a plurality of nodes connected by the links,
Each node represents a biomolecule,
Determination method of control target node. The method according to claim 1, wherein the activity of any node included in the first signaling network represents the amount of activated biomolecules represented by the arbitrary node. The method according to claim 1, wherein the control target preset for the first output node is to increase or decrease the activity of the first output node. The method according to claim 1, wherein the first output node is any one of a node indicating cell death, a node indicating proliferation of a cell, and a node indicating cell metastasis. delete A method for determining a control target node to be controlled for controlling an output node of a signal transmission network, the method comprising:
calculating, by a computing device, an influence on a first output node of each of a plurality of nodes included in a first signal delivery network, and calculating a plurality of influences corresponding to each of the plurality of nodes;
calculating, by the computing device, an SPLO from each of the plurality of nodes to the first output node, and calculating, by the computing device, a plurality of SPLOs corresponding to each of the plurality of nodes;
classifying, by the computing device, the plurality of nodes into a plurality of SPLO groups based on the value of the SPLO; and
determining, by the computing device, as a control target node for controlling the first output node by independently selecting at least some of the nodes belonging to each SPLO group for each of the plurality of SPLO groups;
includes,
The influence of any first node among the plurality of nodes on any output node is the degree of influence of the change in the activity of the first node on the change in activity (∂x _{i ) of the arbitrary output node.} ego,
The change in the activity of the first node is the change in the basal activity of the first node (∂b _j ),
The influence of the first node on the first output node (S _ij ) satisfies Equation 1 ([Equation 1] S _ij = ∂x _i / ∂b _j ),
Determination method of control target node. A method for determining a control target node to be controlled for controlling an output node of a signal transmission network, the method comprising:
For each of the plurality of output nodes included in the first signaling network:
The computing device calculates the influence on the output node of each of the plurality of nodes included in the first signal delivery network, and calculates the plurality of influences corresponding to each of the plurality of nodes,
The computing device calculates an SPLO from each of the plurality of nodes to the output node, and calculates a plurality of SPLOs corresponding to each of the plurality of nodes,
The computing device classifies the plurality of nodes into a plurality of SPLO groups based on the value of the SPLO, and
so that the computing device independently selects at least some of the nodes belonging to each SPLO group for each of the plurality of SPLO groups and determines as a control target node for controlling the output node.
determining a set of control target candidates for each output node; and
a control target node determining step of determining, by the computing device, a final control target node for controlling the first output node based on the intersection of the control target nodes determined for each of the plurality of output nodes;
includes,
The influence of any first node among the plurality of nodes on any output node is the degree of influence of the change in the activity of the first node on the change in activity (∂x _{i ) of the arbitrary output node.} ego,
The change in the activity of the first node is the change in the basal activity of the first node (∂b _j ),
The influence of the first node on the first output node (S _ij ) satisfies Equation 1 ([Equation 1] S _ij = ∂x _i / ∂b _j ),
Determination method of control target node. 10. The method of claim 9, wherein the SPLO from any of the nodes to any of the output nodes connects the any node and the any output node to each other to reach from the any node to the any output node. A method of determining a control target node, indicating the number of links included in the shortest path having the smallest number of links included in each candidate path among one or more candidate paths that can be selected from one or more links. The method of claim 1 , wherein the influence of the first output node of any first node of the plurality of nodes on the first output node is that of the first signaling network from the arbitrary first node to the first output node. A method of determining a control target node, which is calculated using an operation that sums the products of weights assigned to links included in each path in all paths. 12. The method of claim 11,
The weights assigned to the links included in each path are,
obtaining, by the computing device, structural information of the signaling network including information about a source node and a target node connected to each link, and information about a sign of a weight mapped to each link; and
a weight generation step of generating, by the computing device, a value of a weight mapped to each link by using the structure information;
containing,
Which is generated using a method for generating information about the signaling network,
Determination method of control target node. 13. The method of claim 12,
The weight generation step is
generating, by the computing device, an adjacency matrix using the structure information;
generating, by the computing device, a diagonal matrix from the adjacent matrix; and
calculating, by the computing device, a weight value mapped to each link using a weight matrix generated by performing a matrix multiplication operation on the adjacent matrix and the diagonal matrix
containing,
Determination method of control target node. A method for determining a biomolecule that should be a target of a drug in order to control the activity of an output biomolecule, which is a specific biomolecule, in a biomolecule network composed of interactions between biomolecules, the method comprising:
The computing device calculates the influence on the output biomolecule of each of the plurality of biomolecules included in the biomolecule network, and calculates the influence of a plurality of influences corresponding to each of the plurality of biomolecules. Luance calculation step; and
The computing device selects some of the plurality of biomolecules according to a preset control target for the output biomolecule based on the values of the plurality of influences to control the output biomolecule. a drug target biomolecule determination step of determining as a drug target biomolecule for
includes,
The influence of any first biomolecule among the plurality of biomolecules on the output biomolecule is a degree to which a change in the activity of the first biomolecule affects a change in the activity of the output biomolecule;
The change in the activity of the first node is the change in the basal activity of the first node (∂b _j ),
The influence of the first node on the first output node (S _ij ) satisfies Equation 1 ([Equation 1] S _ij = ∂x _i / ∂b _j ),
A method for determining a drug target biomolecule.

Images