Fault-tolerant oriented physical design method for fully programmable valve array biochip

Abstract

The present application relates to a kind of completely programmable valve array biochip fault-tolerant oriented physical design method, comprising the following steps: step S1: component layout constraint is constructed, and based on the dynamic constraint layout algorithm of particle swarm optimization, dynamic fault-tolerant layout is carried out in combination with dynamic fault-tolerant technique, obtains a fault-tolerant layout scheme without component conflict;Step S2: considering reagent input sequence, based on the A* fault-tolerant routing algorithm of best input sequence, dynamic fault-tolerant routing is carried out in combination with dynamic fault-tolerant technique;Step S3: based on the routing optimization strategy of priority and backtracking, fluid conflict occurring in routing is further processed, to obtain the fault-tolerant routing scheme without fluid conflict.The present application generates fault-tolerant FPVA architecture that can avoid all physical failures and potential conflicts, so as to realize a physical design solution scheme that minimizes bioassay completion time and total length of transport path.

Description

Technical Field

[0001] The present invention relates to the technical field of computer-aided design of microfluidic biochips, and in particular to a fault-tolerant and guided physical design method for a fully programmable valve array biochip. Background Art

[0002] As a new generation of flow microfluidic biochips, fully programmable valve array (FPVA) biochips have become a popular biochemical experimental platform, which combines the advantages of traditional flow and digital microfluidic biochips, performs fluid operations based on channel and valve structures, and provides programmability and high flexibility. Figure 1 As shown, the FPVA biochip consists of intersecting flow channels with independently addressable valves arranged in a regular array. By switching the valve states, fluids can be flexibly transported and components dynamically constructed anywhere on the FPVA. The architectural synthesis design process for FPVA biochips is divided into high-level synthesis and physical design. The physical design phase, as the most critical, includes component layout and flow channel routing, defining the chip's fundamental flow channel network and thus impacting its overall performance. Therefore, to improve the quality and efficiency of FPVA biochips, an efficient physical design automation solution is essential. Furthermore, due to environmental or human factors, manufacturing defects such as channel breakage, misalignment, and unintended connections are unavoidable during the manufacturing process. These faults can cause bioassay failures, significantly increasing chip manufacturing costs and even posing a risk. However, no fault-tolerance algorithms have been proposed to address these faults on FPVA biochips. Therefore, to ensure the correct execution of on-chip bioassays, fault-tolerant design is essential. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide a fault-tolerant physical design method for a fully programmable valve array biochip, aiming to solve the above problems.

[0004] To achieve the above object, the present invention adopts the following technical solutions:

[0005] A fault-tolerant physical design method for a fully programmable valve array biochip comprises the following steps:

[0006] Step S1: Construct component layout constraints, and perform dynamic fault-tolerant layout based on a dynamic constraint layout algorithm based on particle swarm optimization and combined with dynamic fault-tolerant technology to obtain a fault-tolerant layout solution without component conflicts.

[0007] Step S2: Considering the reagent input order, the A* fault-tolerant routing algorithm based on the optimal input sequence is combined with dynamic fault-tolerant technology to perform dynamic fault-tolerant routing;

[0008] Step S3: Based on the routing optimization strategy of priority and backtracking, the fluid conflicts occurring during routing are further processed to obtain a fault-tolerant routing solution without fluid conflicts.

[0009] Furthermore, the component layout constraints specifically include:

[0010] A component non-overlap constraint set is introduced to prevent two components from overlapping, thus preventing parallel components from occupying the same unit:

[0011]

[0012]

[0013]

[0014]

[0015]

[0016] in, and Represents operation O i The lower left corner coordinates and upper right corner coordinates of the bound component, K is a huge constant; c k is an auxiliary binary variable, where k = 1, 2, 3, 4, when c k When it is 0, the corresponding constraint holds;

[0017] The component overlap permission is introduced to allow them to overlap, so that the intermediate fluid can proceed to the next step without transportation. The component overlap permission is achieved by adding an auxiliary binary variable c5 in (5):

[0018]

[0019] When the two components are non-concurrent, c5 is set to 1. k , k=1,2,3,4, must all be 1 to allow two devices to overlap.

[0020] Furthermore, the dynamic constraint layout algorithm based on particle swarm optimization is specifically as follows:

[0021] Encode each particle as (pos1,…,pos n ,f); where n is the number of components, pos i Represents component M iThe position information of the component is uniquely determined by the coordinates of the lower left corner and the shape, and f represents the fitness of the particle in the solution space, and

[0022] (1) The fitness function is defined as follows:

[0023] Fitness = a1*OL - a2*A - a3*D - β (7)

[0024] where OL is the number of overlapping units of the parent and child components, A is the chip area occupied by the non-parent and child components, D is the distance from the output fluid to the output port, a1, a2, and a3 are weights that add up to 1, and β is a penalty term for violating the component layout constraints, which is defined as follows:

[0025]

[0026] where K is a very large constant, represents the overlap of concurrent components;

[0027] (2) According to the fitness of the particle, the optimal solution is iteratively searched in the solution space; in each iteration, the particle is updated based on the individual historical optimal solution and the global optimal solution; the update formula of the ith particle P i t is defined as follows:

[0028]

[0029] where M represents the mutation operation, the inertia weight w represents the probability of mutation of the particle, C1 and C2 represent the crossover operation of the particle with the individual historical optimal solution and the global optimal solution, respectively, and the learning factors c1 and c2 represent the probability of crossover of the particle with the individual historical optimal solution and the global optimal solution, respectively;

[0030] (3) First, the velocity is updated, and the update formula is defined as follows:

[0031]

[0032] where r1 is a random number between 0 and 1, and the particle is mutated when it is less than w;

[0033] The mutation operation is to randomly select a component and find a new available unit for it as its new layout position;

[0034] Second, self-cognition learning is performed, and the update formula is defined as follows:

[0035]

[0036] Wherein, r2 is a random number within 0-1, when it is less than c1, the particle crosses with the individual historical optimal solution. The idea of the crossing operation of the particle and the individual historical optimal solution is: keeping the components at the same position of the current particle and the individual historical optimal solution, and randomly selecting a component in the remaining components and finding a new layout position for the component;

[0037] Finally, social experience learning is carried out, and the formula is updated as follows:

[0038]

[0039] Wherein, r3 is a random number within 0-1, when it is less than c2, the particle crosses with the global optimal solution; the idea of the crossing operation of the particle and the global optimal solution is the same as c1.

[0040] Further, the fault-tolerant layout scheme without component conflict includes:

[0041] (1) According to the dual redundancy technology, a new alternative path is constructed using the standby unit;

[0042] (2) According to the unit function conversion technology, the unit occupied by other components that have completed operation is converted into an idle unit, thereby opening up a passable path for the current fluid.

[0043] Further, the A* fault-tolerant routing based on the optimal input sequence is specifically:

[0044] The reagent input sequence is considered, and the minimum biological assay completion time, total length of transportation path and number of fluid conflicts are taken as the target, through formula (13), the new routing cost is calculated to increase the cost of the routing scheme with fluid conflict, which is calculated as:

[0045]

[0046] Wherein, (x p ,y p ) and (x s ,y s ) represent the coordinates of the current unit p and the starting unit s respectively, l represents the unit length through one FPVA unit, and cell(s, p) represents the set of routing units passed from the starting unit s to the current unit p.

[0047] Further, the routing optimization strategy based on priority is specifically:

[0048] The fluid transportation priority is defined,

[0049] Level 1: fluid with earlier operation time;

[0050] Level 2: fluid farther away from the target component;

[0051] Level 3: Fluid in transport state;

[0052] Level 4: Fluids that are stored for longer periods of time;

[0053] When a flow conflict occurs, the lower priority flow is redirected or stored in a backup unit, waiting for the higher priority flow to pass first.

[0054] Furthermore, if a fluid is transported to a cell already occupied by another fluid, the fluid in the cell will pass through first regardless of the priority.

[0055] Furthermore, the backtracking-based routing optimization strategy is specifically as follows: repeatedly executing the following process until the conflict is avoided: when a fluid conflict occurs at a certain moment, backtracking to the previous moment and looking for available backup units to store the critical fluid; at the same time, the remaining fluids continue to be transported according to the fluid transportation priority.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] The present invention generates a fault-tolerant FPVA architecture that can avoid all physical failures and potential conflicts, thereby achieving a physical design solution that minimizes the bioassay completion time and the total length of the transportation path. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is a diagram of the FPVA architecture and its topological connection in one embodiment of the present invention;

[0059] Figure 2 It is a flow chart of the method of the present invention;

[0060] Figure 3 It is a multifunctional unit in one embodiment of the present invention;

[0061] Figure 4 It is a bidirectional redundancy technology in one embodiment of the present invention;

[0062] Figure 5 This is a fault mapping technology in an embodiment of the present invention. DETAILED DESCRIPTION

[0063] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0064] In this embodiment, physical failures on FPVA biochips are defined as failures caused by environmental or mold problems during chip manufacturing and are divided into the following four categories:

[0065] (1) Valve obstruction: due to accidental connection between control channels, air pressure is transmitted to non-designated channel segments, causing the valve of the channel segment to fail to open normally, thereby blocking the transportation of fluid.

[0066] (2) Valve leakage: the valve fails to close normally, causing the fluid to fail to stay in the designated channel segment. The causes of this failure include but are not limited to: 1) due to the breakage of the control channel, unevenness of the control channel wall or misalignment of the control channel and the flow channel, the control channel is blocked, and air pressure cannot reach the designated position; 2) due to accidental connection between control channels, channel wall perforation and other factors, the control channel leaks, and the air pressure reaching the designated position is insufficient; 3) the formed valve loses elasticity or the channel size is mismatched, causing the formed valve to fail to cover and seal the corresponding flow channel segment.

[0067] (3) Flow channel obstruction: due to factors such as breakage of the flow channel, unevenness of the flow channel wall, etc., the flow channel is blocked, and the fluid cannot pass through normally.

[0068] (4) Flow channel leakage: due to factors such as accidental connection between flow channels, channel wall perforation, etc., the flow channel leaks, and the fluid penetrates into other channel segments or is lost.

[0069] In addition to physical failures, component and fluid conflicts in physical design are two other factors that hinder the correct execution of bioassays.

[0070] (1) Component conflict: during the layout phase, multiple components simultaneously occupy the same cell, or there are not enough free cells to simultaneously construct multiple components, which will cause component conflict.

[0071] (2) Fluid conflict: during the routing phase, multiple fluids simultaneously transport to the same cell, or a fluid transports to a cell that has been occupied by another fluid, which will cause fluid conflict.

[0072] In this embodiment, based on the flexibility of FPVA, the present application introduces the concept of multifunctional cells, and defines three types of functional cells, as shown in Figure 3 , including:

[0073] (1) Transport cell: by opening or closing a series of valves on the transport cell, the flow direction of the fluid is guided to form a transport path, and the fluid is transported between the input / output port and the component.

[0074] (2) Storage cell: by closing all peripheral valves on the storage cell, the fluid is temporarily stored therein.

[0075] (3) Mixing cell: by opening the internal valves of the mixing cell and closing its peripheral valves, mixers of different sizes are dynamically constructed, and the fluid is driven to mix therein.

[0076] Based on the concept of multifunctional units, this paper proposes a unit function conversion technology: during the bioassay process, each unit can dynamically convert its function, allowing the fluid to achieve specific functions without additional transportation, such as dynamic reconstruction of components and control of fluid state conversion.

[0077] In this embodiment, based on the unit function conversion technology, the present invention designs a bidirectional redundancy technology, such as Figure 4 As shown, this technology has the following characteristics:

[0078] (1) Adjacent FPVA units serve as backup units for each other, so that there are 2-4 backup units around one FPVA unit.

[0079] (2) Each backup unit can functionally replace any adjacent original unit and can dynamically implement different functions.

[0080] Bidirectional redundancy technology maximizes the use of all idle units on the FPVA architecture and prepares backup resources for each unit without adding any additional cost, thus achieving real-time fault tolerance at the lowest cost.

[0081] In this embodiment, in combination with the above-mentioned fault tolerance technology, the present invention also proposes a fault mapping technology. Taking into account the different types and characteristics of physical faults, they are mapped to corresponding positions on the FPVA topology connection diagram according to the following rules to generate a fault map, such as Figure 5 shown.

[0082] (1) Since the valve blocking fault indicates that the passage section where the valve is located is prohibited, the valve blocking fault is mapped to a fault connection.

[0083] (2) Since the valve leakage fault only affects the two units connected by the valve, and except for this faulty valve, the other valves and all channel sections in the two units are normal, the valve leakage fault is mapped to the merged unit.

[0084] (3) If the fluid passes through a channel section with channel blockage or channel leakage, it may cause fluid loss or contamination. That is, the occurrence of channel blockage and channel leakage will cause the corresponding FPVA unit to be unusable. Therefore, channel blockage and channel leakage are mapped as faulty units.

[0085] refer to Figure 2 The present invention provides a fault-tolerant physical design method for a fully programmable valve array biochip, comprising the following steps:

[0086] Step S1: Construct component layout constraints, and perform dynamic fault-tolerant layout based on a dynamic constraint layout algorithm based on particle swarm optimization and combined with dynamic fault-tolerant technology to obtain a fault-tolerant layout solution without component conflicts.

[0087] Step S2: Considering the reagent input order, the A* fault-tolerant routing algorithm based on the optimal input sequence is combined with dynamic fault-tolerant technology to perform dynamic fault-tolerant routing;

[0088] Step S3: Based on the routing optimization strategy of priority and backtracking, the fluid conflicts occurring during routing are further processed to obtain a fault-tolerant routing solution without fluid conflicts.

[0089] Preferably, in this embodiment, the component layout constraints specifically include:

[0090] A component non-overlap constraint set is introduced to prevent two components from overlapping, thus preventing parallel components from occupying the same unit:

[0091]

[0092]

[0093]

[0094]

[0095]

[0096] in, and Represents operation O i The lower left corner coordinates and upper right corner coordinates of the bound component, K is a huge constant; c k is an auxiliary binary variable, where k = 1, 2, 3, 4, when c k When it is 0, the corresponding constraint holds;

[0097] Since components can be dynamically constructed and reconstructed, when the operation bound to a component is completed, the unit where the component is located is immediately converted into an available unit. Therefore, when two components are not parallel, component overlap permissions are introduced to allow them to overlap, so that the intermediate fluid can proceed to the next operation without transportation. The component overlap permission is implemented by adding an auxiliary binary variable c5 in (5):

[0098]

[0099] When the two components are non-concurrent, c5 is set to 1. k , k=1,2,3,4, must all be 1 to allow two devices to overlap.

[0100] Preferably, in this embodiment, the dynamic constraint layout algorithm based on particle swarm optimization is specifically:

[0101] According to the component position, an encoding scheme is designed in this paper, which encodes each particle as (pos1, …, posn, f). Where n is the number of components, posi represents the position information of component M n i, and f represents the fitness value of the particle. i i i t

[0102] (1) The fitness function is defined as follows:

[0103] Fitness = a1 x OL - a2 x A - a3 x D - β (7)

[0104] Where OL is the number of overlapping units of parent and child components, A is the chip area occupied by non-parent and child components, D is the distance from the output fluid to the output port, a1, a2, a3 are weights that add up to 1, and β is a penalty term for violating component layout constraints, defined as follows:

[0105]

[0106] Where K is a large constant, represents the overlap of concurrent components.

[0107] (2) According to the fitness of the particle, the optimal solution is searched in the solution space iteratively; in each iteration, the particle updates according to the individual historical optimal solution and the global optimal solution; the update formula of the ith particle P i t in the tth iteration is defined as follows:

[0108]

[0109] Where M represents the mutation operation, the inertia weight w represents the probability of mutation of the particle, C1 and C2 represent the crossover operation of the particle with the individual historical optimal solution and the global optimal solution, respectively, and the learning factors c1 and c2 represent the probability of crossover of the particle with the individual historical optimal solution and the global optimal solution, respectively.

[0110] (3) First, update the speed, the update formula is defined as follows:

[0111]

[0112] Where r1 is a random number between 0 and 1, and when it is less than w, the particle undergoes mutation.

[0113] The mutation operation is to randomly select a component and find a new available unit for it as its new layout position.

[0114] Second, self-cognition learning, the update formula is defined as follows:

[0115]

[0116] where r2 is a random number between 0 and 1, and when it is less than c1, the particle crosses with the individual historical optimal solution. The idea of the crossing operation between the particle and the individual historical optimal solution is: keeping the components at the same position of the current particle and the individual historical optimal solution, and randomly selecting a component from the remaining components and finding a new layout position for it;

[0117] Finally, social experience learning is performed, and the formula is updated as follows:

[0118]

[0119] where r3 is a random number between 0 and 1, and when it is less than c2, the particle crosses with the global optimal solution; the idea of the crossing operation between the particle and the global optimal solution is the same as c1.

[0120] In the embodiment, preferably, according to the physical fault mapping diagram, the fault connection and the fault unit are marked as unusable units, if the fluid passes through the unusable unit, the application proposes two wiring fault-tolerant schemes: 1) according to the bidirectional redundancy technology, a new alternative path is constructed using the standby unit thereof, 2) according to the unit function conversion technology, the units occupied by other components that have completed operation are converted into idle units, thereby opening up a passable path for the current fluid.

[0121] In the embodiment, since multiple reagents are transported from the input port to the components, unreasonable reagent input order can cause multiple fluid conflicts to occur. Therefore, the application considers the reagent input order, takes minimizing the bioassay completion time, the total length of the transportation path and the number of fluid conflicts as the goal, and proposes an A* fault-tolerant wiring algorithm based on the optimal input sequence. In addition, the application designs a new wiring cost to increase the cost of the wiring scheme with fluid conflicts, which is calculated as

[0122]

[0123] where (x p ,y p ) and (x s ,y s ) represent the coordinates of the current unit p and the starting unit s respectively, l represents the unit length passing through one FPVA unit, and cell(s, p) represents the set of wiring units passing through from the starting unit s to the current unit p.

[0124] In the embodiment, the priority-based wiring optimization strategy is as follows:

[0125] The fluid transportation priority is defined as:

[0126] Priority (high to low) Fluid type Level 1 Fluids with earlier operation time Level 2 Fluids that are farther away from the target component Level 3 Fluid in transport Level 4 Longer storage time of fluids

[0127] When a fluid conflict occurs, the fluid with lower priority changes direction or is stored in a spare unit, waiting for the fluid with higher priority to pass first. In particular, if a fluid is transported to a unit that has been occupied by another fluid, the fluid in the unit will pass first regardless of priority.

[0128] In this embodiment, a backtracking-based routing optimization strategy is used, specifically: repeatedly execute the following process until conflicts are avoided: when a fluid conflict occurs at a certain time, backtrack to the previous time and find an available spare unit to store the key fluid (i.e., the fluid causing the conflict). At the same time, the remaining fluids continue to be transported according to the fluid transport priority.

[0129] The above only describes the preferred embodiments of the present application, and any changes and modifications made within the scope of the patent application of the present application shall be covered by the present application.

Claims

1. A fault-tolerant physical design method for a fully programmable valve array biochip, characterized in that: The following steps are involved: Step S1: Construct component layout constraints, and perform dynamic fault-tolerant layout based on a dynamic constraint layout algorithm based on particle swarm optimization and combined with dynamic fault-tolerant technology to obtain a fault-tolerant layout solution without component conflicts. Step S2: Considering the reagent input order, the A* fault-tolerant routing algorithm based on the optimal input sequence is combined with dynamic fault-tolerant technology to perform dynamic fault-tolerant routing; Step S3: Based on the priority and backtracking routing optimization strategy, further processing the fluid conflicts that occur during routing to obtain a fault-tolerant routing solution without fluid conflicts; The dynamic constraint layout algorithm based on particle swarm optimization is specifically as follows: Encode each particle as (pos1,…,pos n ,f); where n is the number of components, pos i Represents component M i The position information of the particle is uniquely determined by the coordinates and shape of the lower left corner of the component, and f represents the fitness value of the particle. (1) The fitness function is defined as follows: Fitness=α1×OL-α2×A-α3×D-β (7) Where OL is the number of overlapping units of parent-child components, A is the chip area occupied by non-parent-child components, D is the distance from the output fluid to the output port, α1, α2, and α3 are weights that add up to 1, and β is the penalty term for violating component layout constraints, which is defined as follows: Among them, K is a very large constant. Represents overlap of concurrent components; (2) According to the fitness value of the particle, iteratively search for the optimal solution in the solution space; in each round of iterative search, the particle completes the update based on the individual historical optimal solution and the global optimal solution; in the tth round of iteration, the i-th particle P i t The update formula is defined as follows: P i t =C2(C1(M(P i t-1 ,w),c1)c2) (9) Among them, M represents the mutation operation, the inertia weight w represents the probability of particle mutation, C1 and C2 represent the crossover operation between the particle and the individual historical optimal solution and the global optimal solution, respectively, and the learning factors c1 and c2 represent the probability of the particle crossing the individual historical optimal solution and the global optimal solution, respectively; (3) First, the speed is updated. The update formula is defined as follows: Among them, r1 is a random number between 0 and 1. When it is less than w, the particle will mutate; The mutation operation is: randomly select a component and find a new available unit as its new layout position; Secondly, self-cognitive learning is carried out and the update formula is defined as follows: Where r2 is a random number between 0 and 1. When it is less than c1, the particle crosses with the individual historical optimal solution. The idea of ​​the crossover operation between the particle and the individual historical optimal solution is to retain the components in which the current particle and the individual historical optimal solution are in the same position, and randomly select a component from the remaining components to find a new layout position for it. Finally, social experience learning is carried out and the update formula is defined as follows: Among them, r3 is a random number between 0 and 1. When it is less than c2, the particle crosses with the global optimal solution. The idea of ​​the crossover operation between the particle and the global optimal solution is the same as C1.

2. The fault-tolerant physical design method for a fully programmable valve array biochip according to claim 1, characterized in that: The component layout constraints specifically include: A component non-overlap constraint set is introduced to prevent two components from overlapping, thus preventing parallel components from occupying the same unit: in, and Represents operation O i The lower left corner coordinates and upper right corner coordinates of the bound component, K is a huge constant; c k is an auxiliary binary variable, where k = 1, 2, 3, 4, when c k When it is 0, the corresponding constraint holds; The component overlap permission is introduced to allow them to overlap, so that the intermediate fluid can proceed to the next step without transportation. The component overlap permission is achieved by adding an auxiliary binary variable c5 in (5): When the two components are non-concurrent, c5 is set to 1. k , k=1,2,3,4, must all be 1 to allow two devices to overlap.

3. The fault-tolerant physical design method for a fully programmable valve array biochip according to claim 1, characterized in that: The component-conflict-free fault-tolerant layout solution includes: (1) Using its backup unit to build a new alternative path based on bidirectional redundancy technology; (2) According to the unit function conversion technology, the units occupied by other components that have completed the operation are converted into idle units, thereby opening up a passable path for the current fluid.

4. The fault-tolerant physical design method for a fully programmable valve array biochip according to claim 1, characterized in that: The A* fault-tolerant routing based on the optimal input sequence is specifically as follows: Taking the reagent input order into consideration, with the goal of minimizing the bioassay completion time, the total length of the transport path, and the number of fluid conflicts, the new wiring cost is calculated by formula (13) to increase the cost of the wiring scheme with fluid conflicts, which is calculated as: Among them, (x p ,y p ) and (x s ,y s ) represent the coordinates of the current cell p and the starting cell s, l represents the unit length passing through an FPVA cell, and cell(s,p) represents the set of wiring cells passing from the starting cell s to the current cell p.

5. The fault-tolerant physical design method for a fully programmable valve array biochip according to claim 1, characterized in that: The priority-based routing optimization strategy is specifically as follows: Define fluid transport priorities, Level 1: Fluids with earlier operation time; Level 2: fluid farther from the target component; Level 3: Fluid in transport state; Level 4: Fluids that are stored for longer periods of time; When a flow conflict occurs, the lower priority flow is redirected or stored in a backup unit, waiting for the higher priority flow to pass first.

6. The fault-tolerant physical design method for a fully programmable valve array biochip according to claim 5, characterized in that: If a fluid is transported to a cell already occupied by another fluid, the fluid in the cell will be given priority regardless of its priority.

7. The fault-tolerant physical design method for a fully programmable valve array biochip according to claim 1, characterized in that: The routing optimization strategy based on backtracking is as follows: the following process is repeated until the conflict is avoided: when a fluid conflict occurs at a certain moment, backtrack to the previous moment and find available backup units to store the critical fluid; at the same time, the remaining fluids continue to be transported according to the fluid transportation priority.

Images