CN108510436A - Reconstruction parameter searching method and system in a kind of Ice mapping three-dimensionalreconstruction - Google Patents
Reconstruction parameter searching method and system in a kind of Ice mapping three-dimensionalreconstruction Download PDFInfo
- Publication number
- CN108510436A CN108510436A CN201810267090.9A CN201810267090A CN108510436A CN 108510436 A CN108510436 A CN 108510436A CN 201810267090 A CN201810267090 A CN 201810267090A CN 108510436 A CN108510436 A CN 108510436A
- Authority
- CN
- China
- Prior art keywords
- sampled point
- parameter
- sampled
- dimensionalreconstruction
- likelihood score
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 59
- 238000013507 mapping Methods 0.000 title claims abstract description 23
- 238000005070 sampling Methods 0.000 claims abstract description 39
- 238000009826 distribution Methods 0.000 claims abstract description 32
- 238000002474 experimental method Methods 0.000 claims abstract description 31
- 238000012952 Resampling Methods 0.000 claims abstract description 29
- 238000000342 Monte Carlo simulation Methods 0.000 claims abstract description 12
- 238000007476 Maximum Likelihood Methods 0.000 claims abstract description 10
- 230000008569 process Effects 0.000 claims abstract description 8
- 238000013519 translation Methods 0.000 claims description 6
- 239000002245 particle Substances 0.000 abstract description 11
- 238000005259 measurement Methods 0.000 abstract description 8
- 239000000523 sample Substances 0.000 description 17
- 238000003384 imaging method Methods 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 238000005481 NMR spectroscopy Methods 0.000 description 3
- 241000208340 Araliaceae Species 0.000 description 2
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 2
- 235000003140 Panax quinquefolius Nutrition 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 239000004744 fabric Substances 0.000 description 2
- 230000008014 freezing Effects 0.000 description 2
- 238000007710 freezing Methods 0.000 description 2
- 235000008434 ginseng Nutrition 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000002424 x-ray crystallography Methods 0.000 description 2
- 230000003321 amplification Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008827 biological function Effects 0.000 description 1
- 239000012472 biological sample Substances 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000009510 drug design Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000010894 electron beam technology Methods 0.000 description 1
- 238000001493 electron microscopy Methods 0.000 description 1
- 238000001803 electron scattering Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000005304 joining Methods 0.000 description 1
- 229920002521 macromolecule Polymers 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 102000004169 proteins and genes Human genes 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000007873 sieving Methods 0.000 description 1
Classifications
-
- G06T3/08—
Abstract
Reconstruction parameter searching method and system in a kind of Ice mapping three-dimensionalreconstruction of present invention offer, the method includes：Build higher-dimension parameter space, to each experiment photo, stochastical sampling carried out in the parameter space by Monte-carlo Simulation Method, experiment with computing photo on each sampled point with the likelihood score of setting models；It is more than the sampled point to impose a condition to likelihood score and carries out resampling, generates new stochastical sampling point, and calculate corresponding likelihood score；This resampling process is repeated, until all sampled points converge near the sampled point with maximum likelihood degree；It is described a series of statistical distribution parameter of sampled points after convergence as a kind of statistics of the reconstruction parameter to the experiment photo, and the three-dimensional electronic density map for reconstructing large biological molecule.Individually accurate measurement can be carried out to the defocusing amount parameter that each large biological molecule particle is imaged, and can greatly improve the three-dimensionalreconstruction resolution ratio of large biological molecule.
Description
Technical field
The present invention relates to structure biology technical field, more particularly, to being reconstructed in a kind of Ice mapping three-dimensionalreconstruction
Parameter searching method and system.
Background technology
The microtechnic for using transmission electron microscope observation sample at low temperature is just called freezing electron micrology technology
(cryo-electron microscopy, cryo-EM), abbreviation Ice mapping technology.Ice mapping technology is important structure
Biological study method, it is with other two kinds of technologies：X-ray crystallography (X-ray crystallography) and nuclear magnetic resonance
The basis that (nuclear magnetic resonance, NMR) together forms high resolution structures biological study, is obtaining
The structure of large biological molecule simultaneously discloses particularly important in terms of its function.The basic principle of Ice mapping technology is exactly that sample is frozen
Come and then low temperature is kept to put into inside microscope, biological sample is observed as light source using relevant electron beam, penetrates
After sample and neighbouring ice sheet, the image that lens system is converted to electron scattering signal amplification is recorded on the detector,
Picture signal processing is finally carried out, the three-dimensional structure of sample is obtained.Individual particle Ice mapping Three Dimensional Reconfiguration is freezing electricity
A series of photo of the large biological molecule with homogeneous structural of random orientations of sub- microscope photographing, passes through a set of three-dimensionalreconstruction
Algorithm calculate its high-resolution three-dimension structure.Reconstruct the row that the three-dimensional structure come discloses atom in large biological molecule
Row mode and the pattern of interaction.By the analysis to structure, relevant biological function and inherent mechanism can be explained, it is right
Understand that the basic principle of life, the molecule mechanism of disease and drug design etc. are of great significance.
Three-dimensionalreconstruction needs to obtain the photo shot from each different orientation angle around sample first, then
The three-dimensional structure of sample can be reconstructed.Since the angle of each protein shooting is uncontrollable, three-dimensionalreconstruction algorithm includes two portions
Point, first part is to calculate the three dimensions of every photo the parameter needed for three-dimensionalreconstructions, referred to as reconstruction parameter such as to be orientated, the
Two parts are to carry out three-dimensionalreconstruction according to the reconstruction parameter calculated.It is multiple that the three dimensions of one photo of description is orientated needs
Translation parameters in parameter, including two faces, three spatial orientation angular dimensions, sorting parameter and the relevant parameter of imaging etc..
These reconstruction parameters are accurately measured, are the deciding factors for determining final three-dimensionalreconstruction resolution ratio.However, because the quantity of photo
Very huge, usually in tens of thousands of orders of magnitude to hundreds of thousands, this also means that millions of or even millions parameters need
It accurately to be measured.Meanwhile in the photo for reconstruct, the photo of many low quality even impurity is often adulterated, it is right
The measurement of reconfigurable measurement parameter causes interference, often influences the final resolution ratio of structure determination.Therefore, how accurately to measure
These parameters, and the accuracy of each location parameter is assessed, it is an extremely important problem.
The parameters such as the orientation of every photo for three-dimensionalreconstruction, which may be considered that, to be distributed in a multi-C parameter space,
Each parameter corresponds to a dimension.As described above, current three-dimensionalreconstruction parameter space at least 5 dimensions, including 2 flat
It moves and 3 spatial orientation angles.If it is considered that imaging parameters etc., it is necessary to increase to higher dimension.Based on such multidimensional sky
Between method is described, at present people introduced in Ice mapping three-dimensionalreconstruction many kinds of parameters search method.Wherein most often
The method of grid search, basic principle be in given parameter search range, it is right according to a fixed step-length
All possible parameter is attempted one by one, is eventually found the maximum parameter of possibility as search result；This method is most
It is big the disadvantage is that calculation amount is with the improve of search precision, operand is in exponential increase；For example, in 5 dimension spaces, Mei Gekong
Between search for 10 times, then total volumes of searches be 105It is secondary.If 1 times of search precision raising, each dimensional searches 20 times, then
Calculation amount just becomes original 25=64 times.One flexible method is first to use thicker grid search, then determines one
Then rough parameter area does finer search in this small range again.No matter that grid-search algorithms is all strong
The strong precision dependent on parameter search, and need to determine that the search of starting compensates by the knowledge of some priori.
Also a kind of searching method is the method declined based on gradient, and this method is by estimating gradient near starting point
Change to determine the direction of search, can use less volumes of searches that can be quickly found out optimal parameter, accordingly, with respect to grid
Searching method, great advantage are that speed is fast, but this method can only solve the problems, such as local optimum, are only capable of carrying out local search.
And gradient method, with the increase of dimension, search reliability is remarkably decreased.The method that grid combination gradient declines also frequent quilt
Using doing global search using the coarse grid in covering global parameter space, it is accurate that the method then declined using gradient does part
Search, but either grid search or gradient search, all can not make assessment to the search reliability of each parameter.
Invention content
The present invention provides a kind of a kind of Ice mapping three for overcoming the above problem or solving the above problems at least partly
Dimension reconstruct in reconstruction parameter searching method and system, solve in the prior art can not spatial parameter search speed it is slow, reliability
It is low, and the problem of assessment can not be made to the search reliability of each parameter.
According to an aspect of the present invention, reconstruction parameter searching method in a kind of Ice mapping three-dimensionalreconstruction is provided, including：
Build parameter space, to each experiment photo, by Monte-carlo Simulation Method in the parameter space into
Row stochastical sampling, experiment with computing photo on each sampled point with the initial likelihood score of setting models；
It is more than the sampled point to impose a condition to initial likelihood score and carries out resampling, generate new sampled point, and calculates corresponding
Likelihood score；
The resampling process is repeated, until the distribution mean square deviation of all sampled points no longer reduces；
It is retouched the statistical distribution parameter of the sampled point after convergence as a kind of statistics of the reconstruction parameter to the experiment photo
It states, and the three-dimensional electronic density map for reconstructing large biological molecule.
Preferably, structure parameter space specifically includes：
Distance is built, the distance is passed through into two translation parameters descriptions of x and y；It is empty to build gyrator
Between, the rotation subspace is passed through into a unit quaternion q description；Defocus vector subspace is built, by defocus quantum sky
Between pass through the variation proportionality coefficient ζ descriptions of defocusing amount；Configuration state subspace is built, the configuration state subspace is led to
The integer number μ of configuration state belonging to a description is crossed to describe；
The distance, rotation subspace, defocus vector subspace and configuration state subspace are combined into parameter sky
Between { x, y, q, ζ, μ }.
Preferably, the likelihood score for obtaining each sampled point and experiment photo specifically includes：
Three-dimensional object of reference is projected by the parameter of sampled point, calculates the likelihood score between projection and experiment photo.
It is specifically included preferably, being more than the corresponding sampled point of projection to impose a condition to likelihood score and carrying out resampling：
Using the likelihood score as the weight of sampled point, height sequence is carried out to sampled point according to weight, will be sorted forward
N number of sampled point carry out resampling, i.e., regenerate multiple sampled points centered on each former sampled point, come subsequent weight
Lower sampled point will be removed, and ensure that sum of the total number of sample points before and after resampling is constant.
Preferably, further including after being more than the sampled point to impose a condition progress resampling to likelihood score：
Every time after sampling, sampling point distributions situation is counted, is carried out based on the sampling point distributions mean square deviation
Next round resampling.
Preferably, including up to the distribution mean square deviation of all sampled points no longer reduces tool：
So that the likelihood score of all sampled points is restrained, if resampling can not make sampled point converge to smaller region, makes institute
There is sampled point to converge near the sampled point with maximum likelihood degree.
Space parameter search system in a kind of Ice mapping three-dimensionalreconstruction, including：
Sampling module, for each experiment photo, carried out in parameter space by Monte-carlo Simulation Method with
Machine samples, and obtains multiple sampled points；
Search module projects three-dimensional with reference to object for the parameter based on each sampled point, obtains each projection
With the likelihood score of experiment photo；
Loop module, for being more than the corresponding sampled point of projection to impose a condition to likelihood score, as the sampling module is sent out
It send resampling to instruct, and when all sampled points converge to the corresponding sampled point of the projection with maximum likelihood degree, sends and stop
Sampling instruction.
Reconstructed module, the reconstruction parameter information for being reflected according to sampled point carries out three-dimensionalreconstruction, sampled point point
The inverse of cloth mean square deviation randomly selects N number of or whole sampled points and participates in three-dimensionalreconstruction by this weight as weight.
Preferably, further including confidence level module, the confidence level module is used for after each round samples, statistics
Sampled point and calculates the mean square deviation that sampled point is distributed in parameter space in the distribution situation of parameter space.
Preferably, further including weight computation module, the weight computation module in all sampled points for converging to
After the corresponding sampled point of projection with maximum likelihood degree, the inverse of the mean square deviation of the sampling point distributions is calculated, and returned
One change is handled, by the weight reciprocal as corresponding experiment photo.
Reconstruction parameter searching method and system in a kind of Ice mapping three-dimensionalreconstruction of present invention proposition, in Ice mapping three-dimensional
Estimate parameter using the important sampling algorithm of particle filter class in reconstruct, the method based on stochastical sampling carries out parameter Estimation,
It realizes and the confidence level of single parameter Estimation in Ice mapping three-dimensionalreconstruction is measured, to improve the Shandong of higher-dimension parameter Estimation
Stick can be searched for more effectively and be orientated relevant parameter and progress two and three dimensions classification, moreover it is possible to joined to the defocusing amount of imaging
Number carry out local searches to greatly improve the three-dimensionalreconstruction resolution ratio of some samples, meanwhile, but also to imaging when sample thickness
Defocusing amount measurement error caused by degree and inclination has good adaptability, and atom definition is made more easily to obtain.
Description of the drawings
Fig. 1 is according to space parameter searching method flow diagram in the Ice mapping three-dimensionalreconstruction of the embodiment of the present invention.
Specific implementation mode
With reference to the accompanying drawings and examples, the specific implementation mode of the present invention is described in further detail.Implement below
Example is not limited to the scope of the present invention for illustrating the present invention.
As shown in Figure 1, reconstruction parameter searching method in a kind of Ice mapping three-dimensionalreconstruction is shown in figure, including：
Build parameter space, to each experiment photo, by Monte-carlo Simulation Method in the parameter space into
Row stochastical sampling obtains each sampled point and the likelihood score of experiment photo；Specifically, for a given three-dimensional reference object
With an experiment photo, three-dimensional object of reference is projected according to a set of given parameter, the phase between projection and experiment photo
It is described with likelihood score (likelihood) like degree.To each such parameter, its corresponding likelihood score can be calculated.
Resampling is carried out more than the sampled point to impose a condition to likelihood score, until all sampled points are converged to maximum seemingly
The region so spent is described the distributed constant of sampled point as a kind of statistics for describing measured reconstruction parameter.By joining
The parameter for determining a set of global optimum is searched in number space, and calculates the confidence level of each parameter.It is so-called " optimal ", refer to
It is to have maximum between the structural information under this set parameter entrained by calculated three-dimensionalreconstruction electron-density map and experiment photo
Likelihood score.
The basis of model three-dimensionalreconstruction is central section theorem, and the critical issue in restructuring procedure is how to determine each
The Space Angle (orientation determination) of grain image.Most models reconstruct and optimization algorithm is all based on throwing
Shadow matches the alternative manner of (projection matching).It is exactly briefly, first with coarse 3 d structure model, into
The image that row projection is referred to, and experiment particle image are compared, and update dimensional orientation parameter according to result, then structure
Make new three-dimensional structure, the dimensional orientation amendment to experimental image forms the process of iteration, until convergence just obtain it is final
Threedimensional model.During three-dimensionalreconstruction, each orientation has translation, rotation, in addition spatial orientation angle, comes to five freedom
Degree, it is next that each image at least needs five parameters three-dimensional structure could be reconstructed.Only every experiment picture is all determined
After corresponding orientation, three-dimensionalreconstruction could be carried out.
Monte Carlo method (Monte Carlo method), also referred to as statistical simulation methods, refer to using random number (or more
Common pseudo random number) come the method that solves many computational problems.The course of solving questions of Monte Carlo method can be attributed to three
Key step：Construction or description probabilistic process；It realizes to be distributed from known probability and sample；Establish various estimators.Particle filter
The thought of (Particle Filter, PE) is based on monte carlo method (Monte Carlo methods), it is to utilize particle
Collect to indicate probability, it can be on any type of state-space model.Its core concept is by from Posterior probability distribution
The stochastic regime particle of middle extraction expresses its distribution, is a kind of sequence importance sampling method (Sequential Importance
Sampling).In simple terms, particle filter method refers to close to probability by finding one group of random sample propagated in state space
It spends function and carries out approximation, integral operation is replaced with sample average, to obtain the process of state minimum variance distribution.Here sample
This refers to particle, and any type of probability density distribution can be approached as sample size N → ∝.
Specifically, in the present embodiment, this parameter space is divided into multiple subspaces, including distance, rotation
Rotor space, defocus vector subspace and configuration state subspace.Distance is by two translation parameters descriptions of x and y；Gyrator
Space is described by a unit quaternion q；Defocus vector subspace is described by the variation proportionality coefficient ζ of a defocusing amount；Knot
Structure subspace method is a discrete parameter, and the number of affiliated configuration state is described by an integer μ.Build parameter space
It specifically includes：
Distance is built, the distance is passed through into two translation parameters descriptions of x and y；It is empty to build gyrator
Between, the rotation subspace is passed through into a unit quaternion q description；Defocus vector subspace is built, by defocus quantum sky
Between pass through the variation proportionality coefficient ζ descriptions of defocusing amount；Configuration state subspace is built, the configuration state subspace is led to
The integer number μ of configuration state belonging to a description is crossed to describe；
The distance, rotation subspace, defocus vector subspace and configuration state subspace are combined into parameter sky
Between { x, y, q, ζ, μ }.On this basis, this method, which can not only be searched for more effectively, is orientated two peacekeepings of relevant parameter and progress
Three-dimensional categorisation, moreover it is possible to which local search is carried out to the defocusing amount parameter of imaging.The search of defocusing amount parameter can be greatly improved
The three-dimensionalreconstruction resolution ratio of sample, especially when high resolution is in 3 angstroms.Meanwhile but also to imaging when sample thickness and
Defocusing amount measurement error has good adaptability caused by inclination, and atom definition is made more easily to obtain.
Sampling point distributions meansquaredeviationσ is initialized, calculates the weight of sampled point, and the sampled point of high weight is equal according to distribution
Variances sigma is split, and removes low weight sampled point, the sum of sampled point is constant, specifically, in the present embodiment, being more than to likelihood score
The corresponding sampled point of projection of setting condition carries out resampling and specifically includes：
Using the likelihood score as the weight of sampled point, height sequence is carried out to sampled point according to weight, will be sorted forward
N number of sampled point carry out resampling, i.e., regenerate multiple sampled points centered on each former sampled point, come subsequent weight
Lower sampled point will be removed, and ensure that sum of the total number of sample points before and after resampling is constant.
The resampling process is repeated, until the distribution mean square deviation of all sampled points no longer reduces, makes all sampled points
Likelihood score is restrained, if resampling can not make sampled point converge to smaller region, all sampled points is made to converge to maximum
Near the sampled point of likelihood score.Search of the estimation through excessively taking turns of entire parameter, iteration operation.Initial ranging uses Monte Carlo
The mode of (Monte Carlo) simulation stochastical sampling in parameter space.Then calculate the likelihood score of each sampled point, and by its
Weight as the sampled point.Each sampled point will be split according to the height of its weight, referred to as resampling.Each weight
High sampled point will be split as multiple sampled points, and be dispersed near original sampled point.Set total sampled point
Number remains unchanged before and after resampling.The lower sampled point of those weight ratios, will be removed.It reruns this sampling and again
The sampled point of sampling process, these fixed quantities will gradually converge near the sampled point (parameter) with maximum likelihood degree.
Specifically, during above-mentioned iterative estimate, the first round is initial estimation wheel, and a large amount of sampled points are evenly distributed
In given parameter space.With the later several rounds on the basis of first round, by resampling, the ginseng of global optimum is gradually converged to
Near number.When each round terminates, the distribution situation of sampled point is counted, by the square of the sampling point distributions of statistics
Foundation of the difference as next round resampling.Statistical distribution situation of the sampled point in parameter space, reflects the general of parameter measurement
Rate density function describes the confidence level of this parameter Estimation.A final wheel calculates the inverse of the mean square deviation of sampling point distributions,
After normalization, it is taken as the weight of this photo, for adjusting its contribution in three-dimensionalreconstruction.
Weight adjustment in three-dimensionalreconstruction is embodied in two aspects.One is that the pixel value of experiment photo is directly multiplied by power
Weight.The other is every photo is used repeatedly all in three-dimensionalreconstruction, but corresponding parameter is all different when each use.These
Parameter is random acquirement, and has identical distribution with the sampled point of an above-mentioned final wheel.
This method, which realizes, measures the confidence level of single parameter Estimation in Ice mapping three-dimensionalreconstruction, to improve height
Tie up the robustness of parameter Estimation.The confidence level of parameter evaluation is further as weight factor, for adjusting current photo to most
The contribution of whole three-dimensionalreconstruction.Tolerance of this method to bad photo greatly improved in this function, reduces the difficulty of optical sieving
Degree.Whether to the refine of defocusing amount parameter, or to the high tolerance of bad photo, be all conducive to applying for automatized three-dimensional reconstruct
Row, it is often more important that provide guarantee for the extensive high-throughput structure of biological macromolecule measurement of futurity industry.
Reconstruction parameter search system in a kind of Ice mapping three-dimensionalreconstruction is additionally provided in the present embodiment, including：
Sampling module, for each experiment photo, carried out in parameter space by Monte-carlo Simulation Method with
Machine samples, and obtains multiple sampled points；
Search module projects three-dimensional with reference to object for the parameter based on each sampled point, obtains each projection
With the likelihood score of experiment photo；
Loop module, for being more than the corresponding sampled point of projection to impose a condition to likelihood score, as the sampling module is sent out
It send resampling to instruct, and when all sampled points converge to the corresponding sampled point of the projection with maximum likelihood degree, sends and stop
Sampling instruction.
Reconstructed module, the reconstruction parameter information for being reflected according to sampled point carries out three-dimensionalreconstruction, sampled point point
The inverse of cloth mean square deviation randomly selects N number of or whole sampled points and participates in three-dimensionalreconstruction by this weight as weight.
In the present embodiment, further include confidence level module, the confidence level module is used for after each round samples, system
Sampled point is counted in the distribution situation of parameter space, and calculates the mean square deviation that sampled point is distributed in parameter space.
In the present embodiment, further include weight computation module, the weight computation module is used to restrain in all sampled points
To the corresponding sampled point of projection with maximum likelihood degree, the inverse of the mean square deviation of the sampling point distributions is calculated, and carry out
Normalized, by the weight reciprocal as corresponding experiment photo.
In conclusion space parameter searching method and system in a kind of Ice mapping three-dimensionalreconstruction of present invention proposition, cold
Freeze in Electronic Speculum three-dimensionalreconstruction and estimate parameter using the important sampling algorithm of particle filter class, the method based on stochastical sampling carries out
Parameter Estimation is realized and is measured the confidence level of single parameter Estimation in Ice mapping three-dimensionalreconstruction, to improve higher-dimension ginseng
The robustness of number estimation can be searched for more effectively and be orientated relevant parameter and progress two and three dimensions classification, moreover it is possible to imaging
Defocusing amount parameter carry out local search, to the search of defocusing amount parameter can greatly improve some samples three-dimensionalreconstruction resolution
Rate, meanwhile, but also imaging when sample thickness and inclination caused by defocusing amount measurement error have good adaptability, make
Atom definition more easily obtains.
Finally, method of the invention is only preferable embodiment, is not intended to limit the scope of the present invention.It is all
Within the spirit and principles in the present invention, any modification, equivalent replacement, improvement and so on should be included in the protection of the present invention
Within the scope of.
Claims (10)
1. reconstruction parameter searching method in a kind of Ice mapping three-dimensionalreconstruction, which is characterized in that including：
Build parameter space, to each experiment photo, carried out in the parameter space by Monte-carlo Simulation Method with
Machine sample, experiment with computing photo on each sampled point with the initial likelihood score of setting models；
It is more than the sampled point to impose a condition to initial likelihood score and carries out resampling, generate new sampled point, and calculates accordingly seemingly
So degree；
The resampling process is repeated, until the distribution mean square deviation of all sampled points no longer reduces；
It is described the statistical distribution parameter of the sampled point after convergence as a kind of statistics of the reconstruction parameter to the experiment photo, and
Three-dimensional electronic density map for reconstructing large biological molecule.
2. according to the method described in claim 1, it is characterized in that, structure parameter space specifically includes：
Distance is built, the distance is passed through into two translation parameters descriptions of x and y；Structure rotation subspace, will
The rotation subspace passes through a unit quaternion q description；Defocus vector subspace is built, the defocus vector subspace is passed through
The variation proportionality coefficient ζ descriptions of one defocusing amount；Configuration state subspace is built, the configuration state subspace is passed through one
The integer number μ of affiliated configuration state is described to describe；
By the distance, rotation subspace, defocus vector subspace and configuration state subspace be combined into parameter space x,
y,q,ζ,μ}。
3. according to the method described in claim 1, it is characterized in that, obtain experiment photo on each sampled point with setting models
Likelihood score specifically include：
Given three-dimensional object of reference is projected by the parameter of sampled point, calculates the likelihood score between projection and experiment photo.
4. according to the method described in claim 1, adopted it is characterized in that, being more than the sampled point to impose a condition to likelihood score again
Sample specifically includes：
Using the likelihood score as the weight of sampled point, height sequence is carried out to sampled point according to weight, will be sorted forward N number of
Sampled point carries out resampling as former sampled point, and multiple sampled points, and resampling are reacquired centered on each former sampled point
Total number of sample points afterwards is identical as the total number of sample points before resampling.
5. according to the method described in claim 1, adopted it is characterized in that, being more than the sampled point to impose a condition to likelihood score again
Further include after sample：
Every time after sampling, sampling point distributions situation is counted, is carried out based on the sampling point distributions mean square deviation next
Take turns resampling.
6. according to the method described in claim 1, it is characterized in that, until the distribution mean square deviation of all sampled points no longer reduces tool
Including：
So that the likelihood score of all sampled points is restrained, if resampling can not make sampled point converge to smaller region, makes all adopt
Sampling point converges near the sampled point with maximum likelihood degree.
7. according to the method described in claim 1, it is characterized in that, the statistical distribution parameter of sampled point is used as to the experiment photo
Reconstruction parameter a kind of statistics description further include：
The inverse of the mean square deviation of the sampling point distributions is calculated, and is normalized, it will be described reciprocal as corresponding experiment
The weight of photo；Randomly selected from sampled point it is N number of or whole, by identical weight participate in three-dimensionalreconstruction.
8. reconstruction parameter search system in a kind of Ice mapping three-dimensionalreconstruction, which is characterized in that including：
Sampling module, for each experiment photo, being adopted at random in parameter space by Monte-carlo Simulation Method
Sample obtains multiple sampled points；
Search module projects with reference to object three-dimensional for the parameter based on each sampled point, obtain each projection in fact
The likelihood score of passport control examination of passports piece；
Loop module sends weight for being more than the corresponding sampled point of projection to impose a condition to likelihood score to the sampling module
Sampling instruction, and when all sampled points converge to the corresponding sampled point of the projection with maximum likelihood degree, send and stop sampling
Instruction.
Reconstructed module, the reconstruction parameter information for being reflected according to sampled point carries out three-dimensionalreconstruction, and sampling point distributions are equal
The inverse of variance randomly selects N number of or whole sampled points and participates in three-dimensionalreconstruction by this weight as weight.
9. system according to claim 8, which is characterized in that further include confidence level module, the confidence level module is used for
After each round samples, statistic sampling point parameter space distribution situation, and calculate sampled point parameter space be distributed
Mean square deviation.
10. system according to claim 8, which is characterized in that further include weight computation module, the weight computation module
After converging to the corresponding sampled point of the projection with maximum likelihood degree in all sampled points, the sampling point distributions are calculated
The inverse of mean square deviation, and be normalized, by the weight reciprocal as corresponding experiment photo.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810267090.9A CN108510436B (en) | 2018-03-28 | 2018-03-28 | Method and system for searching reconstruction parameters in three-dimensional reconstruction of cryoelectron microscope |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810267090.9A CN108510436B (en) | 2018-03-28 | 2018-03-28 | Method and system for searching reconstruction parameters in three-dimensional reconstruction of cryoelectron microscope |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108510436A true CN108510436A (en) | 2018-09-07 |
| CN108510436B CN108510436B (en) | 2020-09-15 |
Family
ID=63379162
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810267090.9A Active CN108510436B (en) | 2018-03-28 | 2018-03-28 | Method and system for searching reconstruction parameters in three-dimensional reconstruction of cryoelectron microscope |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108510436B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109444842A (en) * | 2019-01-04 | 2019-03-08 | 北京环境特性研究所 | A kind of electromagnetic characteristic of scattering data reconstruction method and device |
| CN110032761A (en) * | 2019-03-07 | 2019-07-19 | 浙江工业大学 | A kind of classification method of electron cryo-microscopy individual particle imaging data |
| CN110490883A (en) * | 2019-08-22 | 2019-11-22 | 南京信易达计算技术有限公司 | A kind of electron cryo-microscopy data analyzed pattern system and method based on web |
| CN116580158A (en) * | 2023-06-15 | 2023-08-11 | 北京大学 | Method, apparatus and storage medium for generating simulated image of frozen electron tomography |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1287766A (en) * | 1998-01-20 | 2001-03-14 | 新型转换器有限公司 | Active acoustic devices comprising panel members |
| CN101051385A (en) * | 2006-04-07 | 2007-10-10 | 欧姆龙株式会社 | Tracking method and device for special shooted objects and tracking method and device for aspect parts |
| CN101447092A (en) * | 2008-12-24 | 2009-06-03 | 苏州和君科技发展有限公司 | Method for accelerating volume rendering during post treatment of MicroCT original image |
| CN103733133A (en) * | 2011-08-15 | 2014-04-16 | 佳能株式会社 | Image capturing apparatus, method of controlling the same and program |
| CN104077769A (en) * | 2014-06-06 | 2014-10-01 | 华南理工大学 | Error matching point pair removing algorithm in image registration |
| CN107493124A (en) * | 2017-08-09 | 2017-12-19 | 深圳先进技术研究院 | A kind of beamforming algorithm of multiple antennas microwave wireless charging |
-
2018
- 2018-03-28 CN CN201810267090.9A patent/CN108510436B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1287766A (en) * | 1998-01-20 | 2001-03-14 | 新型转换器有限公司 | Active acoustic devices comprising panel members |
| CN101051385A (en) * | 2006-04-07 | 2007-10-10 | 欧姆龙株式会社 | Tracking method and device for special shooted objects and tracking method and device for aspect parts |
| CN101447092A (en) * | 2008-12-24 | 2009-06-03 | 苏州和君科技发展有限公司 | Method for accelerating volume rendering during post treatment of MicroCT original image |
| CN103733133A (en) * | 2011-08-15 | 2014-04-16 | 佳能株式会社 | Image capturing apparatus, method of controlling the same and program |
| CN104077769A (en) * | 2014-06-06 | 2014-10-01 | 华南理工大学 | Error matching point pair removing algorithm in image registration |
| CN107493124A (en) * | 2017-08-09 | 2017-12-19 | 深圳先进技术研究院 | A kind of beamforming algorithm of multiple antennas microwave wireless charging |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109444842A (en) * | 2019-01-04 | 2019-03-08 | 北京环境特性研究所 | A kind of electromagnetic characteristic of scattering data reconstruction method and device |
| CN109444842B (en) * | 2019-01-04 | 2020-09-25 | 北京环境特性研究所 | Target electromagnetic scattering characteristic data reconstruction method and device |
| CN110032761A (en) * | 2019-03-07 | 2019-07-19 | 浙江工业大学 | A kind of classification method of electron cryo-microscopy individual particle imaging data |
| CN110032761B (en) * | 2019-03-07 | 2023-07-25 | 浙江工业大学 | Classification method for single-particle imaging data of frozen electron microscope |
| CN110490883A (en) * | 2019-08-22 | 2019-11-22 | 南京信易达计算技术有限公司 | A kind of electron cryo-microscopy data analyzed pattern system and method based on web |
| CN116580158A (en) * | 2023-06-15 | 2023-08-11 | 北京大学 | Method, apparatus and storage medium for generating simulated image of frozen electron tomography |
| CN116580158B (en) * | 2023-06-15 | 2023-11-17 | 北京大学 | Method, apparatus and storage medium for generating simulated image of frozen electron tomography |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108510436B (en) | 2020-09-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108510436A (en) | Reconstruction parameter searching method and system in a kind of Ice mapping three-dimensionalreconstruction | |
| Barath et al. | MAGSAC: marginalizing sample consensus | |
| Heymann | Guidelines for using Bsoft for high resolution reconstruction and validation of biomolecular structures from electron micrographs | |
| Strandlie et al. | Track and vertex reconstruction: From classical to adaptive methods | |
| Taha et al. | An efficient algorithm for calculating the exact Hausdorff distance | |
| Lawrence et al. | Transform-based backprojection for volume reconstruction of large format electron microscope tilt series | |
| CN109087342A (en) | A kind of three-dimensional point cloud global registration method and system based on characteristic matching | |
| CN112567231A (en) | Inspection of articles by dynamically selecting projection angles | |
| Bortel et al. | Common arc method for diffraction pattern orientation | |
| Feuerstein et al. | Reconstruction of 3-D histology images by simultaneous deformable registration | |
| Shi et al. | Evaluation of the performance of classification algorithms for XFEL single-particle imaging data | |
| Brubaker et al. | Building proteins in a day: Efficient 3D molecular reconstruction | |
| WO2012008512A1 (en) | Image processing method, image processing system, and x-ray computed tomography system | |
| Li | Imaging of orientation and geometry in microstructures: development and applications of high energy x-ray diffraction microscopy | |
| US9857163B2 (en) | Parametric control of object scanning | |
| Ge et al. | A point cloud registration method combining enhanced particle swarm optimization and iterative closest point method | |
| CN110108302B (en) | Method for improving atom group polishing precision | |
| Mousavi et al. | Evaluating tie points distribution, multiplicity and number on the accuracy of uav photogrammetry blocks | |
| Yuce | Novel computational technique for determining depth using the Bees Algorithm and blind image deconvolution | |
| Magri et al. | Bending the doming effect in structure from motion reconstructions through bundle adjustment | |
| Shen et al. | A point cloud registration algorithm for the fusion of virtual and real maintainability test prototypes | |
| WO2023004558A1 (en) | Neural implicit function for end-to-end reconstruction of dynamic cryo-em structures | |
| Yao et al. | Robust surface registration using N-points approximate congruent sets | |
| Lyckegaard | Development of tomographic reconstruction methods in materials science with focus on advanced scanning methods | |
| Öfverstedt et al. | Automated correlation and averaging of three-dimensional reconstructions obtained by electron tomography |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |