Physical Design Method of Flow Layers Based on Discrete Particle Swarm under Continuous Microfluidic Biochips

Abstract

The present invention relates to a physical design method of flow layers based on discrete particle swarm under a continuous microfluidic biochip, which comprises the following steps: Step S1: Construct a particle coding scheme based on sequence pairs according to the characteristics of the physical design problem of the continuous microfluidic biochip; Step S2: Search the solution space of the physical design of the flow layers of the continuous microfluidic biochip based on the particle update strategy; Step S3: Generate a design scheme that meets the preset requirements according to the particle coding scheme and the solution space obtained in Step S2. The present invention can effectively improve the efficiency of the physical design of flow layers.

Description

Technical Field

[0001] The present invention relates to the technical field of computer-aided design of continuous microfluidic biochips, and particularly relates to a physical design method of flow layers based on discrete particle swarm under a continuous microfluidic biochip. Background Art

[0002] In recent years, continuous microfluidic biochips have attracted great research interest from scientific researchers. The continuous microfluidic biochip technology integrates basic operations such as sample preparation, reaction, separation, and detection in the biological, chemical, and medical analysis processes onto a chip with a micron scale, and automatically completes the entire process of experimental analysis. Due to its great potential in the fields of biology, chemistry, medicine, etc., it has now developed into a brand-new research field that intersects multiple disciplines such as biology, chemistry, medicine, fluid, electronics, materials, and machinery, and is applied in biological analysis technologies (such as DNA detection, enzyme-linked immunosorbent assay, enzyme-linked immunosorbent assay), clinical diagnosis (such as antibody-based diagnosis, cancer diagnosis), food safety detection (such as detection of pesticide residues, veterinary drug residues, heavy metal residues), etc.

[0003] In the existing work, the integer linear programming method is used to implement the actual operations of fluid transportation and removal of excess fluid (or waste liquid). However, this method has a high complexity and cannot meet the needs of large-scale integration of continuous microfluidic biochips. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a physical design method of flow layers based on discrete particle swarm under a continuous microfluidic biochip, which can effectively improve the efficiency of physical design of flow layers.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions:

[0006] The present invention has the following beneficial effects compared with the prior art:

[0007] A physical design method of flow layers based on discrete particle swarm under a continuous microfluidic biochip includes the following steps:

[0008] Step S1: According to the characteristics of the physical design problem of the continuous microfluidic biochip, construct a particle coding scheme based on sequence pairs;

[0009] Step S2: Based on the particle update strategy, search the solution space of the physical design of the flow layers of the continuous microfluidic biochip;

[0010] Step S3: According to the particle coding scheme and the solution space obtained in step S2, generate a design scheme that meets the preset requirements.

[0011] Further, the specific steps of step S1 are as follows: Obtain relevant information of component devices, flow path information, parallel execution flow paths, and the positions of excess liquid and waste liquid, and then use the representation method based on sequence pairs to encode the particles to obtain a particle encoding scheme.

[0012] Further, the particle update strategy is specifically as follows:

[0013] The update formula is as follows:

[0014]

[0015] Among them, the first part represents the influence of the current state of the particle, which is expressed as:

[0016]

[0017] Among them, r represents a random number in the interval [0, 1], and f(X i (k)) represents the update of the particle velocity. If the generated random number is less than ω, perform the operation of f(X i (k)); otherwise, keep the original particle;

[0018] The second part is the learning of the particle individual to itself, which is expressed as:

[0019]

[0020] Among them, r represents a random number in the interval [0, 1], and g(E i (k), P i (k)) represents that the particle learns towards its own optimal solution P i (k), retain the same device positions in the sequence pair in the individual historical optimal solution, randomly exchange the different positions. If the generated random number is less than c1, perform the operation of g(E i (k), P i (k)); otherwise, keep the original particle, and the result is the result of the first part;

[0021] The third part is the adjustment of the particle individual according to the global optimal position of the group, which is expressed as:

[0022]

[0023] Among them, r represents a random number in the interval [0, 1], and h(Q i (k), G(k)) represents that the particle learns towards the global optimal solution G(k). The same as the previous part, if the generated random number is less than c2, perform the operation of h(Q i(k), G(k)) operation; otherwise, keep the original particles, and the result is the result of the previous part.

[0024] Further, the f(X i (k)) operation is implemented through the mutation operation of the particles, as described in Formula (5)

[0025]

[0026] The specific method is as follows: (1): f1(X i (k)) represents the swap operation: randomly select two devices within a sequence pair and swap their positions, keeping the other positions unchanged; (2): f2(X i (k)) represents the reverse operation: randomly select two devices within a sequence pair, reverse the devices in between, and place them back between the two positions; (3): f3(X i (k)) represents the insertion operation: randomly select two devices within a sequence pair and insert the selected device into the position of the other device.

[0027] Further, in the particle update strategy, the calculation formula of the fitness value function F of each particle is expressed as:

[0028] F = α × N P + β × N I + γ × L C Formula (8)

[0029] In the above Formula (8), N P represents the number of ports of the biochip under this layout, and N I represents the number of flow channel intersection points generated under this layout; L C represents the length of the flow channel.

[0030] Further, the specific step S3 is as follows: adopt the A*-based routing algorithm to obtain the physical design solution corresponding to each particle, calculate the fitness value of each particle, after completion, update the particle fitness value and learn from the individual historical optimal solution and the global optimal solution until the iteration ends and the result is output, and obtain the design scheme that meets the preset requirements.

[0031] The present invention can effectively improve the physical design efficiency of the flow layer. Brief Description of the Drawings

[0032] Figure 1 is the flowchart of the particle swarm algorithm in an embodiment of the present invention;

[0033] Figure 2 is an example of the representation method based on sequence pairs in an embodiment of the present invention;

[0034] Figure 3It is an example of particle mutation operation in an embodiment of the present invention;

[0035] Figure 4 It is an example of a particle learning individual optimal solution in an embodiment of the present invention;

[0036] Figure 5 It is an example of a particle learning global optimal solution in an embodiment of the present invention;

[0037] Figure 6 It is an example of a flow path without considering the removal of excess fluid and waste liquid in an embodiment of the present invention;

[0038] Figure 7 It is an example of a flow path considering the removal of excess fluid and waste liquid in an embodiment of the present invention. Detailed implementation manners

[0039] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0040] Please refer to Figure 1 , the present invention provides a flow layer physical design method based on discrete particle swarm for a continuous microfluidic biochip, including the following steps:

[0041] Step S1: Construct a particle encoding scheme based on sequence pairs according to the characteristics of the physical design problem of the continuous microfluidic biochip;

[0042] Step S2: Search the solution space of the flow layer physical design of the continuous microfluidic biochip based on the particle update strategy;

[0043] Step S3: Generate a design scheme that meets the preset requirements according to the particle encoding scheme and the solution space obtained in step S2.

[0044] In this embodiment, based on the discrete particle swarm algorithm, it is assumed that in a D-dimensional space, there is a population Pop = {X1,..., X i ,..., X m} composed of m initial particles, and each particle has two attributes: its own speed and position information. Among them, the position of the i-th particle is denoted as and its speed is denoted as The historical optimal position of this particle is denoted as The global optimal position of the population is denoted as G = {G1, G2,..., G D}. Therefore, in the d-th dimensional space of the i-th particle, its speed and position can be algorithmically expressed as:

[0045]

[0046]

[0047] In formulas (1) and (2), k represents the generation of iterative evolution; w represents the inertia factor, which is used to adjust the search ability for the solution space; c1 and c2 represent the learning factors, which adjust the maximum step size; r1 and r2 represent random numbers in the interval [0, 1], which are used to enhance the randomness of the search. As Figure 1 shown is the flowchart of the particle swarm algorithm.

[0048] In this implementation, step S1 is specifically: obtaining the relevant information of the component devices, the flow path information, the flow paths for parallel execution, and the positions of the surplus liquid and waste liquid, and then using the representation method based on sequence pairs to encode the particles to obtain the particle encoding scheme.

[0049] As Figure 2 shown, there is currently a set U = {a, b, c, d, e} of component devices, and the currently generated sequence pair is (d1, d2) = (acbed, acebd), and the explanation is as follows: (1) If in the sequence pair, there is a device b behind a, then on the chip, device b can be placed on the left side of a; (2) If in the sequence pair, there is a component device a in front of b in d1 and a component device b in front of a in d2, on the chip, device a can be placed on the upper side of b.

[0050] Preferably, in this embodiment, the particle update strategy is specifically:

[0051] The update formula is as follows:

[0052]

[0053] Among them, the first part represents the influence of the current state of the particle, which is expressed as:

[0054]

[0055] Among them, r represents a random number in the interval [0, 1], f(X i (k)) represents the update of the particle velocity. If the generated random number is less than ω, perform the operation of f(X i (k)); otherwise, keep the original particle;

[0056] The second part is the learning of the particle individual to itself, which is expressed as:

[0057]

[0058] Among them, r represents a random number in the interval [0, 1], g(E i (k), P i (k)) represents the particle moving towards its own optimal solution P i(k) Learn, retain the same device positions within the sequence pair in the individual's historical optimal solution, randomly exchange different positions. If the generated random number is less than c1, perform the operation of g(E i (k), P i (k)); otherwise, keep the original particle, and the result is the result of the first part;

[0059] The third part It is the adjustment of the particle individual according to the global optimal position of the group, expressed as:

[0060]

[0061] Among them, r represents a random number in the interval [0, 1], and h(Q i (k), G(k)) represents that the particle learns towards the global optimal solution G(k). Similar to the previous part, if the generated random number is less than c2, perform the operation of h(Q i (k), G(k)); otherwise, keep the original particle, and the result is the result of the previous part. The operation is as Figure 5 shown.

[0062] Preferably, the operation of f(X i (k)) is implemented through the mutation operation of the particle, as described in formula (5)

[0063]

[0064] The specific method is as follows: (1): f1(X i (k)) represents the exchange operation: randomly select two devices within a sequence pair and exchange their positions, keeping the remaining positions unchanged; (2): f2(X i (k)) represents the reverse order operation: randomly select two devices within a sequence pair, reverse the devices between them, and place them back between the two positions; (3): f3(X i (k)) represents the insertion operation: randomly select two devices within a sequence pair and insert the selected device into the position of another device.

[0065] In this embodiment, in the discrete particle swarm algorithm, the calculation formula of the fitness value function F of each particle is expressed as:

[0066] F = α × N P + β × N I + γ × L C Equation (8)

[0067] In the above equation (8), N P represents the number of ports of the biochip in this layout, and N I represents the number of flow channel intersection points generated in this layout; L CRepresents the length of the flow channel.

[0068] In this embodiment, a fast flow channel routing algorithm is designed based on the A* algorithm. As can be seen from the above, each particle corresponds to a specific layout solution. Using this algorithm, we can quickly obtain the physical design solution corresponding to each particle and calculate the fitness value of each particle.

[0069] It is necessary to meet the needs of actual fluid operation. During the construction of the flow channel, if there is experimental fluid that is no longer needed, it needs to be discharged out of the chip through an independent channel in a timely manner. Assume there is a flow path without the removal of excess fluid (or waste liquid), such as Figure 6 shown.

[0070] During the construction of the flow channel, if there is experimental fluid that is no longer needed, it needs to be discharged out of the chip through an independent channel in a timely manner. If in Figure 6 there is excess fluid at both ends of the mixer after the experimental fluid is mixed, it should be discharged in a timely manner. Independent waste liquid ports should be constructed at both ends of the mixer for timely discharge; if there is excess liquid after heating by the heater in the next step and new experimental fluid needs to be injected, an independent flow channel should be constructed and the corresponding flow ports or waste liquid ports should be arranged at this time. As Figure 7 shown.

[0071] In this embodiment, the flow channel routing is performed using the A* algorithm after the particle update is completed. The routing length is given priority and used as the cost function.

[0072] The A* algorithm selects the node with the minimum routing cost (i.e., Cost(g)) from the priority queue each time as the next node to be traversed. Two tables are used to represent the nodes to be traversed and those that have been estimated using the heuristic function, and the nodes that have been traversed and the routing is successful, denoted as the open table and the close table. The A* algorithm plans the routing path by continuously maintaining the open table and the close table. The steps of the A* algorithm are as follows.

[0073]

[0074] The steps of the flow channel routing algorithm are as follows.

[0075]

[0076]

[0077] Under this problem, the cost function for calculating the flow channel length cost is as follows:

[0078] Cost(g) = G(g) + H(g) Equation (9)

[0079] In the above formula (9), Cost(g) represents the routing length of the flow channel, G(g) represents the path length from the input port to g, and H(g) represents the estimated path length from g to the output port.

[0080] The above are only the preferred embodiments of the present invention, and all equivalent changes and modifications made according to the scope of the patent application of the present invention shall fall within the scope of the present invention.

Claims

1. A physical design method of flow layers based on discrete particle swarm under a continuous microfluidic biochip, characterized in that, Including the following steps: Step S1: Construct a particle coding scheme based on sequence pairs according to the characteristics of the physical design problem of the continuous microfluidic biochip; Step S2: Search the solution space of the physical design of the flow layer of the continuous microfluidic biochip based on the particle update strategy; Step S3: Generate a design scheme that meets the preset requirements according to the particle coding scheme and the solution space obtained in Step S2; The specific content of Step S1 is: Obtain the relevant information of the component devices, flow path information, parallel flow paths, and the positions of excess liquid and waste liquid, and then use the sequence pair-based representation method to encode the particles to obtain a particle coding scheme; The particle update strategy is specifically as follows: The update formula is as follows: Among them, the first part represents the influence of the current state of the particle, expressed as: where r represents a random number within the interval [0, 1], and f(X i (k)) represents the update of the particle velocity. If the generated random number is less than ω, perform the operation of f(X i (k)); otherwise, keep the original particle. The second part It represents the learning of a particle individual about itself, expressed as: Among them, r represents a random number in the interval [0, 1], g(E i (k), P i (k)) means that the particle learns towards its own optimal solution P i (k), retains the same device positions within the sequence pair in the individual historical optimal solution, randomly exchanges different positions, and if the generated random number is less than c1, execute g(E i (k), P i (k)) operation; otherwise, keep the original particle, and the result is the result of the first part; Part Three It means that the particle individuals are adjusted according to the global optimal position of the group, which is expressed as: where r represents a random number in the interval [0, 1], and h(Q i (k), G(k)) represents that the particle learns towards the global optimal solution G(k). Similar to the previous part, if the generated random number is less than c2, perform the operation h(Q i (k), G(k)); otherwise, keep the original particle, and the result is the result of the previous part.

2. The physical design method of flow layers based on discrete particle swarm under a continuous microfluidic biochip according to claim 1, characterized in that, The said f(X i (k)) operation is realized through the mutation operation of particles: The specific methods are as follows: (1): f1(X i (k)) represents a swap operation: randomly select two devices within a sequence pair, swap their positions, and keep the remaining positions unchanged; (2): f2(X i (k)) represents a reverse operation: randomly select two devices within a sequence pair, reverse the devices between them, and place them back between the two positions; (3): f3(X i (k)) represents an insertion operation: randomly select two devices within a sequence pair, and insert the selected device into the position of the other device.

3. The physical design method of flow layers based on discrete particle swarm under a continuous microfluidic biochip according to claim 1, characterized in that, In the particle update strategy, the calculation formula of the fitness value function F of each particle is expressed as: F = α × N P + β × N I + γ × L C where N P represents the number of ports of the biochip in the current layout, and N I represents the number of flow channel intersections generated in the current layout; L C represents the length of the flow channel.

4. The physical design method of flow layers based on discrete particle swarm under a continuous microfluidic biochip according to claim 1, characterized in that, The specific content of Step S3 is: Use the A*-based routing algorithm to obtain the physical design solution corresponding to each particle, calculate the fitness value of each particle, and after completion, update the particle fitness value and learn from the individual historical optimal solution and the global optimal solution until the iteration ends and the result is output to obtain a design scheme that meets the preset requirements.

Images