Embodiment
fig. 1for the flow process of the arrangement of components embodiment of the method one of biochemical reaction pick-up unit on micro chip of the present invention
figure,
fig. 2for the flow process of the arrangement of components embodiment of the method one of biochemical reaction pick-up unit on micro chip of the present invention is illustrated
figure,
as Fig. 1with
fig. 2shown in, on micro chip of the present invention, the arrangement of components method of biochemical reaction pick-up unit comprises:
S101, produce N number of assemblies physical location layout and generate probability distribution parameter corresponding to described assemblies physical location layout; Described biochemical reaction assembly includes but not limited to input or output reservoir, diluting reaction groove (raceway groove); Described router layout includes but not limited to the wiring method being arranged at the raceway groove between reservoir for conducting biochemical reaction reagent or generation biochemical reaction; Described arrangement of components method can be applied in and include but not limited on biochip;
Preferably, described generation N number of assemblies physical location layout (sample) and the probability distribution parameter generating described assemblies physical location layout corresponding comprise:
Random generation assemblies physical location layout of N number of first time and generate first time probability density parameter corresponding to described assemblies physical location layout;
Or,
N number of (i+1) sub-assembly physical location layout is produced according to the probability distribution parameter of the i-th sub-assembly physical location layout after renewal, preferably, comprising: the probability distribution parameter according to the i-th sub-assembly physical location layout after renewal produces N number of (i+1) sub-assembly physical location layout by Monte Carlo simulation.
S102, calculate biochemical reaction deadline of each described assemblies physical location layout; During concrete calculating, the biochemical reaction deadline of each assemblies physical location layout and the funtcional relationship of assemblies physical location layout can represent with the probability distribution parameter of the biochemical reaction deadline of each assemblies physical location layout and assemblies physical location layout; For simplified characterization, below by the probability distribution parameter of the probability distribution parameter of the assemblies physical location layout of biochemical reaction deadline function referred to as assemblies physical location layout; Preferably, the biochemical reaction deadline of each described assemblies physical location layout of described calculating comprises:
The biochemical reaction deadline in multiple router layouts corresponding to each described assemblies physical location layout is calculated respectively according to labyrinth algorithm;
Choose the biochemical reaction deadline of minimum biochemical reaction deadline as described assemblies physical location layout, choose the connection scheme that router layout corresponding to minimum biochemical reaction deadline is each assembly in described assemblies physical location layout simultaneously;
Preferably, illustrate how to use labyrinth algorithm to calculate the biochemical reaction deadline below in conjunction with a concrete assembly layout example:
Comprehensive and physical synthesis two parts are formed by framework for the design of biochip.Framework comprehensively achieves the distribution of limited Resources on Chip to various basic operation, the one-to-one relationship of each assembly (such as mixer and well heater etc.) in each basic operation and actual chips
as Fig. 2shown in.Whole biochip is made up of a series of assembly and micro-valve.Assembly completes the various operations needed for biochemical test, and micro-valve controls reagent in biochip, presses given route transmission.Physical synthesis comprises the layout of each assembly on chip and connects the raceway groove wiring of each assembly.
Framework comprehensive (architectural synthesis) is for determining the one by one mapping of each operation to concrete chip assembly.This comprehensively can be realized by a simple greedy algorithm.This greedy algorithm is according to biochemical test flow process
figuretopological sorting, limited assembly is distributed one by one.A simple framework Comprehensive example
as Fig. 7shown in.According to topological sorting, operation 1,2,4,5,7,8,10,11 does not have preorder to operate, and can give assembly on sheet.Because the resource of this chip limits (only having 6 to input reservoir), operation 1,2,4,5,7,8 are assigned to input 1,2,3,4,5,6 respectively, complete in the 1st time cycle.Operation 10,11 is assigned to input 5 and 6, completes in the 2nd time cycle.Operation 3,6 and 9 is assigned to diluter 1,2 and 3 respectively and completes, and in the cycle at this moment, three diluters on sheet are all used.In the time cycle 3, operation 17,18 and 19 is assigned to output 1,2 and 3 respectively.Now diluter resource is released, and operation 12 is able to carry out in dilution 3.After this Resourse Distribute by that analogy.When Resources on Chip is not enough to meet the current all operations that can carry out, the current executable operation of a part will be delayed to subsequent time period and perform.
as Fig. 4shown in, after biochip layout is determined, we need to have the device of sequential relationship to connect with raceway groove.Between determining device the process of line and VLSI wiring process closely similar, unique difference is: the line in VLSI circuit can not intersect, and the line in biochip allows to intersect.When intersection occurs, the conduction of reagent can not be carried out at infall simultaneously.This means the prolongation testing the deadline.Therefore, raceway groove intersection be should give and avoided.
Maze routing algorithm in VLSI wiring can by the wiring problem solved in biochip.For every a pair interconnective device, the shortest path that maze routing algorithm can find wiring number of crossovers minimum.This algorithm from starting point outside propagation labeled wiring cost, until the every bit of the surrounding of target is labeled.In a certain iteration, each neighbor node of the node that last iteration was accessed is labeled.Simple maze routing algorithm (maze routing algorithm) example
as Fig. 4shown in.
In this instance, input 5 to need to be connected with dilution 3.Existing connection comprises input 1 and dilution 1, dilution 1 and dilution 3.The same color of mark each time in iteration represents.In the first iteration, the position of input about 5 is marked as 1, represents that the cost routing to this from input 5 is 1.In second time iteration, outside propagation from the position of input about 5,
in figure3 position marks are 2.In order to avoid raceway groove intersects, being routed in the nodes of locations cost occupied by other raceway grooves is 2.Such as input the position below 1, the cost propagating into this from input 5 is 2+2=4.After mark completes, can be recalled by impact point and find optimal path.The lowest costs arriving dilution 3 is 7.From then on beginning is put, each lowest costs node finding vicinity.?
fig. 4example in, cost is the lowest costs adjacent node of the node of 7 is 6.By that analogy, until arrive input 5.Final path is by red line mark.
Briefly introduce biochip time series analysis below:
The time series analysis of biochip and framework are comprehensive very similar.Be with the comprehensive difference of framework, time series analysis needs to consider time of channel conduction reagent and raceway groove timesharing and with the delay brought.Example
as Fig. 4in, the raceway groove that raceway groove and the input 5-of input 1-dilution 1 dilute 3 intersects.Therefore, these two raceway grooves cannot cycle operation at one time.This causes input 5 and input 6 needs to move to for the second time cycle from the cycle very first time carrying out.Therefore input 5 postpones with the operation after input 6.
Time series analysis can be realized by a simple greedy algorithm.Seemingly comprehensive with framework, each operation is by flow process
figuretopological sorting process successively.The deadline of each operation is defined by following formula:
T
prevdeadline the latest in preorder operation, T
currthe running time of current operation, T
ddue to raceway groove timesharing and with the delay brought.Last Operation Definition completed deadline of whole biochemical test.
S103, to choose and the probability distribution parameter of the assemblies physical location layout that k biochemical reaction deadline the shortest is corresponding obtains the probability distribution parameter after upgrading before upgrading;
Preferably, to choose described in and the probability distribution parameter of the assemblies physical location layout that k biochemical reaction deadline the shortest is corresponding obtains the probability distribution parameter after upgrading and comprises before upgrading:
Choose the probability distribution parameter of assemblies physical location layout corresponding to front k the shortest biochemical reaction deadline;
According to the probability distribution parameter that the condition minimizing the biochemical reaction deadline by Cross-Entropy Algorithm obtains, the described k chosen a probability distribution parameter is upgraded; Preferably, comprising:
S1031, according to the biochemical reaction deadline minimum condition (1) obtaining the first intermediate function l → 0:
S1032, obtain the second intermediate function according to the condition of the first intermediate function l → 0
with approximate function g
xx condition (2) that the cross entropy (Kullback – Leibler Diffenrence, KL divergence) of () is minimum:
S1033, the probability distribution parameter v ' determined under described cross entropy minimal condition:
S1034, probability distribution parameter v v ' to be upgraded;
Wherein, f
x(x; V) represent that the biochemical reaction deadline is to assemblies physical location layout x (or x
i) function, referred to as f (X), v represents the probability distribution parameter that assemblies physical location layout x is corresponding, g
x(x
i) represent f
x(x; V) approximate function, γ represents the minimum value of biochemical reaction deadline, and P represents probability, I|
f (X)≤γ=1 represents indicator function, I|
f (X)≤γ=1, I|
f (X) > γ=0, N represents the assemblies physical location layout x of generation
inumber;
In order to the implementation of the described method of clearer explanation, below we provide the detailed derivation of S1031 ~ S1034:
For the objective function f (X) be defined on x ∈ X, all x ∈ X are equal
independent.Probability density function f is followed in the distribution of definition x ∈ X
x(x; V), wherein finite dimensional vector v is f
x(x; V) parameter.The target of cross-entropy method determines whether there is constant γ to make f (X)≤γ become small probability event, that is:
l=P(f(X)≤γ)=E[I
{f(X)≤γ}]=∫I
{f(X)≤γ}f
x(x;v) (2);
Because we Water demand l becomes the situation of small probability, we introduce another probability density function g (x), for all x,
the definition of application g (x), l can be expressed as:
Due to X
1... X
nall mutual for g (x)
independentrandom vector, the importance sampling of l can be estimated as:
We need to find a specific g (x), make l ' minimum.The density of stochastic variable X in f (X)≤γ situation is expressed as:
We need to make the Kullback – Leibler difference between g and g' minimum, namely minimize following formula:
We need to find a specific v that-∫ g ' (x) lng (x) dx is minimized, and also namely minimize-∫ g ' (x) lnf (x:v) dx, namely maximize following formula
Also be
max
vD(v)=max
vE
uI
{f(X)≤γ}lnf(x;v) (9);
Importance sampling is utilized to obtain,
Wherein w is arbitrary parameter, is
likelihood ratios.Therefore, v ' can be estimated by following formula
Present stochastic variable x ∈ X obeys probability density function f (x; W).
For solving biochip layout, we have proposed the algorithm based on cross-entropy method.This algorithm from a biochip initial layout iteration optimization, until this layout cannot be optimized further.For the sake of simplicity, primeval life chip is logically divided into N (N>=N
0) individual grid.Wherein N
0for the device populations on biochip, it comprehensively can be obtained by framework.A PCR initial layout
as Fig. 5shown in.By
figurevisible, each device takies separately a net.
Sample in this algorithm refers in particular to the place-exchange of two different components.Such as by input 1 and input 2 switches.The effect that device exchanges characterizes by exchanging mark.Exchange the effect of mark for assessment of current exchange, the test deadline linear inverse ratio that it and current arrangements cause.For exchanging each time, we complete cloth line generalization and time stimulatiom according to current biochip layout, obtain the deadline of whole biochemical test.Different allocation plans can produce different route plannings, and route planning affects the deadline.A good device exchanges and is conducive to reducing the test deadline.
The exchange of every a pair device is selected to obey particular probability distribution.We have sampled normal distribution in algorithm realization.In each iteration, our Stochastic choice n sample (device exchange).For each sample, according to its distribution generation random number, if this number is greater than constant threshold, this sample is selected.For the sample that each is chosen, we are to it
new productraw layout application route planning and time series analysis.After obtaining exchanging mark, the probability distribution that this device exchanges upgrades according to exchange mark.We select best k from all samples chosen, and upgrade their probability distribution, namely increase the average of normal distribution and reduction variance.This will cause the exchange of these devices more to have an opportunity selected in following iteration.Before this iteration terminates, best sample is performed.
new productraw layout becomes the basic layout of next iteration, and all devices exchange and produce in this layout.
Before first time, iteration started, the device that PCR tests sample exchanges probability distribution
as Fig. 6shown in.Namely often pair of exchange has identical chance selected.After an iteration, each probability distribution exchanging combination
as Fig. 7shown in.If k=4, k best sample has been updated probability distribution, makes them in next iteration, have larger probability selected.
The iterative process of this algorithm is continued until cannot optimize distribution further.Each the exchange combination chosen all can not produce in time stimulatiom tests the deadline faster.This represents that this algorithm cannot continue to optimize existing layout.We think existing layout close to optimum.
Illustrate below after providing concrete probability density parameter function, according to the renewal result v ' (3) of the probability distribution parameter that S1031 ~ S1034 provides, obtain the concrete mode of v ' renewal is as described below:
Preferably, described probability density parameter is Multi-dimensional Gaussian distribution parameter vector, and the probability distribution parameter that the condition that described basis minimizes the biochemical reaction deadline by Cross-Entropy Algorithm obtains carries out renewal to the described k chosen a probability distribution parameter and comprises:
Physical location parameter in the probability distribution parameter of the assemblies physical location layout corresponding the biochemical reaction deadline the shortest for front k is set to the mean value of the physical location parameter in the probability distribution parameter of assemblies physical location layout corresponding to front k biochemical reaction deadline the shortest, and before reducing the assemblies physical location layout that k biochemical reaction deadline the shortest is corresponding probability distribution parameter in the value of physical location variance parameter;
S104, produce new sample according to the probability distribution parameter after described renewal;
S105, repeat m described generation layout, described computing time, described in choose and the probability distribution parameter of updated components physical location layout until
new productthe probability distribution parameter of raw assemblies physical location layout provides the assemblies physical location layout determined; Wherein, N>1,1<k<N, m>=1; Here pass through, the result that such iteration is repeatedly optimized is.
On micro chip of the present invention, the arrangement of components method of biochemical reaction pick-up unit is by representing the physical location layout variable probability distribution of biochemical reaction assembly to be optimized, minimize the probability distribution that the reaction deadline optimizes described assembly layout variable again, and then upgrade layout sample set according to the reaction time under this assemblies physical location layout, and this optimizing process that iterates is until probability distribution corresponding to layout sample set provides the layout determined, achieve the appropriate design to assemblies physical position and access path thereof, thus the time of whole biochemical reaction testing process can be reduced.
fig. 1for the flow process of the arrangement of components system embodiment one of biochemical reaction pick-up unit on micro chip of the present invention
figure,
fig. 2for the flow process of the arrangement of components system embodiment one of biochemical reaction pick-up unit on micro chip of the present invention is illustrated
figure,
as Fig. 1with
fig. 2shown in, on micro chip of the present invention, the arrangement of components system of biochemical reaction pick-up unit comprises:
Location layout's module 21, probability distribution parameter corresponding to described assemblies physical location layout is generated for generation of N number of assemblies physical location layout, call result that router layout module complete biochemical deadline calculates and to choose accordingly and the probability distribution parameter of the assemblies physical location layout that k biochemical reaction deadline the shortest is corresponding obtains the probability distribution parameter after renewal before upgrading, new sample is produced according to the probability distribution parameter after described renewal, with repeat m described generation layout, the described biochemical reaction deadline result of calculation calling router layout module, described choose and updated components physical location layout probability distribution parameter until
new productthe probability distribution parameter of raw assemblies physical location layout provides the assemblies physical location layout determined, wherein, N>1,1<k<N, m>=1,
Router layout module 22, for calculating the biochemical reaction deadline of each described assemblies physical location layout.
Described biochemical reaction assembly includes but not limited to input or output reservoir, diluting reaction groove (raceway groove); Described router layout includes but not limited to the wiring method being arranged at the raceway groove between reservoir for conducting biochemical reaction reagent or generation biochemical reaction; Described arrangement of components method can be applied in and include but not limited on biochip;
Preferably, described location layout module 21 specifically for:
Random generation assemblies physical location layout of N number of first time and generate first time probability density parameter corresponding to described assemblies physical location layout, or produce N number of (i+1) sub-assembly physical location layout according to the probability distribution parameter of the i-th sub-assembly physical location layout after renewal, choose the probability distribution parameter of assemblies physical location layout corresponding to front k the shortest biochemical reaction deadline, according to the probability distribution parameter that the condition minimizing the biochemical reaction deadline by Cross-Entropy Algorithm obtains, the described k chosen a probability distribution parameter is upgraded, N number of (i+1) sub-assembly physical location layout is produced by Monte Carlo simulation with according to the probability distribution parameter of the i-th sub-assembly physical location layout after renewal.
Preferably, described location layout module 21 specifically for:
The probability distribution parameter obtained according to the condition minimizing the biochemical reaction deadline by Cross-Entropy Algorithm carries out renewal to the described k chosen a probability distribution parameter and comprises:
According to the biochemical reaction deadline minimum condition (4) obtaining the first intermediate function l → 0:
Condition according to the first intermediate function l → 0 obtains the second intermediate function
with approximate function g
xx condition (7) that the cross entropy (Kullback – Leibler Diffenrence, KL divergence) of () is minimum:
Determine the probability distribution parameter v ' (11) under described cross entropy minimal condition:
Upgrade with to probability distribution parameter v v ';
Wherein, f
x(x; V) represent that the biochemical reaction deadline is to assemblies physical location layout x (or x
i) function, referred to as f (X), v represents the probability distribution parameter that assemblies physical location layout x is corresponding, g
x(x
i) represent f
x(x; V) approximate function, γ represents the minimum value of biochemical reaction deadline, and P represents probability, I|
f (X)≤γ=1 represents indicator function, I|
f (X)≤γ=1, I|
f (X) > γ=0, N represents the assemblies physical location layout x of generation
inumber.
Preferably, described router layout module 22 is specifically for calculating the biochemical reaction deadline in multiple router layouts corresponding to each described assemblies physical location layout according to labyrinth algorithm respectively;
Accordingly, described location layout module 21, specifically for choosing the biochemical reaction deadline of minimum biochemical reaction deadline as described assemblies physical location layout, chooses the connection scheme that router layout corresponding to minimum biochemical reaction deadline is each assembly in described assemblies physical location layout simultaneously.
Last it is noted that above each embodiment is only in order to illustrate technical scheme of the present invention, be not intended to limit; Although with reference to foregoing embodiments to invention has been detailed description, those of ordinary skill in the art is to be understood that: it still can be modified to the technical scheme described in foregoing embodiments, or carries out equivalent replacement to wherein some or all of technical characteristic; And these amendments or replacement, do not make the essence of appropriate technical solution depart from the scope of various embodiments of the present invention technical scheme.