CN108510436B - Method and system for searching reconstruction parameters in three-dimensional reconstruction of cryoelectron microscope - Google Patents
Method and system for searching reconstruction parameters in three-dimensional reconstruction of cryoelectron microscope Download PDFInfo
- Publication number
- CN108510436B CN108510436B CN201810267090.9A CN201810267090A CN108510436B CN 108510436 B CN108510436 B CN 108510436B CN 201810267090 A CN201810267090 A CN 201810267090A CN 108510436 B CN108510436 B CN 108510436B
- Authority
- CN
- China
- Prior art keywords
- sampling points
- sampling
- resampling
- subspace
- likelihood
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 67
- 238000005070 sampling Methods 0.000 claims abstract description 162
- 238000009826 distribution Methods 0.000 claims abstract description 46
- 238000012952 Resampling Methods 0.000 claims abstract description 41
- 230000008569 process Effects 0.000 claims abstract description 13
- 238000007476 Maximum Likelihood Methods 0.000 claims abstract description 12
- 238000000342 Monte Carlo simulation Methods 0.000 claims abstract description 10
- 238000013519 translation Methods 0.000 claims description 22
- 230000008859 change Effects 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000012163 sequencing technique Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 2
- 238000002474 experimental method Methods 0.000 claims 2
- 239000002245 particle Substances 0.000 abstract description 11
- 239000000523 sample Substances 0.000 description 16
- 230000014616 translation Effects 0.000 description 14
- 238000005259 measurement Methods 0.000 description 10
- 238000003384 imaging method Methods 0.000 description 8
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000005057 refrigeration Methods 0.000 description 3
- 238000005481 NMR spectroscopy Methods 0.000 description 2
- 238000001493 electron microscopy Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000008014 freezing Effects 0.000 description 2
- 238000007710 freezing Methods 0.000 description 2
- 238000011478 gradient descent method Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 229920002521 macromolecule Polymers 0.000 description 2
- 230000001172 regenerating effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000002424 x-ray crystallography Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 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
- 230000001427 coherent effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 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
- 238000010894 electron beam technology Methods 0.000 description 1
- 238000001803 electron scattering Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000000386 microscopy Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009456 molecular mechanism Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000001902 propagating 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
- 230000009467 reduction Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
Images
Classifications
-
- G06T3/08—
Abstract
The invention provides a method and a system for searching reconstruction parameters in three-dimensional reconstruction of a cryoelectron microscope, wherein the method comprises the following steps: constructing a high-dimensional parameter space, randomly sampling each experimental photo in the parameter space by a Monte Carlo simulation method, and calculating the likelihood of the experimental photo on each sampling point and a given model; resampling the sampling points with the likelihood degree larger than the set conditions to generate new random sampling points, and calculating the corresponding likelihood degree; repeating the resampling process until all sampling points converge to the vicinity of the sampling point with the maximum likelihood; and taking the statistical distribution parameters of the converged series of sampling points as a statistical description of the reconstruction parameters of the experimental picture, and using the statistical description to reconstruct a three-dimensional electron density map of the biomacromolecule. The defocus parameter of each biomacromolecule particle image can be independently and accurately measured, and the three-dimensional reconstruction resolution of biomacromolecules can be greatly improved.
Description
Technical Field
The invention relates to the technical field of structure biology, in particular to a method and a system for searching reconstruction parameters in three-dimensional reconstruction of a cryoelectron microscope.
Background
The microscopic technique for observing a sample at low temperature by using a transmission electron microscope is called cryo-electron microscopy (cryo-EM), which is called cryo-electron microscopy for short. The cryoelectron microscopy is an important structural biology research method, which is similar to the other two technologies: x-ray crystallography (X-ray crystallography) and Nuclear Magnetic Resonance (NMR) together form the basis of high resolution structural biology research, and are of great importance in obtaining the structure of biological macromolecules and revealing their function. The basic principle of the cryo-electron microscope technology is to freeze a sample, keep the sample at a low temperature, place the sample in a microscope, observe the biological sample by using a coherent electron beam as a light source, convert an electron scattering signal into an amplified image after the sample and a nearby ice layer penetrate through the sample, record the amplified image on a detector by using a lens system, and finally perform image signal processing to obtain the three-dimensional structure of the sample. The single-particle freezing electron microscope three-dimensional reconstruction technology is characterized in that a series of randomly oriented biological macromolecule photos with uniform structures, which are shot by a freezing electron microscope, are calculated to obtain high-resolution three-dimensional structures through a set of three-dimensional reconstruction algorithm. The reconstructed three-dimensional structure reveals the arrangement of atoms in the biomacromolecule and the mode of interaction. By analyzing the structure, the related biological function and the internal mechanism can be explained, and the method has important significance for understanding the basic principle of life, the molecular mechanism of diseases, drug design and the like.
Three-dimensional reconstruction requires first obtaining photographs taken from various different directional angles around the sample before reconstructing the three-dimensional structure of the sample. Because the shooting angle of each protein is not controllable, the three-dimensional reconstruction algorithm comprises two parts, wherein the first part is used for calculating parameters required by three-dimensional reconstruction such as three-dimensional space orientation of each picture and is called reconstruction parameters, and the second part is used for carrying out three-dimensional reconstruction according to the calculated reconstruction parameters. Describing the three-dimensional spatial orientation of a photograph requires a number of parameters including two in-plane translation parameters, three spatial orientation angle parameters, classification parameters, and imaging related parameters, among others. Accurate measurement of these reconstruction parameters is a determining factor in determining the resolution of the final three-dimensional reconstruction. However, because the number of photographs is very large, typically on the order of tens to hundreds of thousands, this means that there are millions or even tens of millions of parameters that need to be accurately determined. Meanwhile, in the photos used for reconstruction, many photos with low quality and even impurities are often doped, which causes interference to the measurement of the reconstruction measurement parameters and often influences the final resolution of the structure determination. Therefore, how to accurately measure these parameters and to evaluate the accuracy of each measured parameter is a very important issue.
For threeParameters such as orientation of each photo reconstructed by dimension can be considered to be distributed in a multi-dimensional parameter space, with each parameter corresponding to a dimension. As mentioned above, the current three-dimensional reconstruction parameter space has at least 5 dimensions, including 2 translations and 3 spatial orientation angles. If imaging parameters, etc., are taken into account, an increase to higher dimensions is required. Based on such a multi-dimensional space description method, people have introduced various parameter search methods in three-dimensional reconstruction of a cryoelectron microscope at present. The most common method is a grid search method, and the basic principle is that in a given parameter search range, all possible parameters are tried one by one according to a fixed step length, and finally, the parameter with the highest possibility is found out to be used as a search result; the biggest defect of the method is that the calculation amount increases exponentially along with the improvement of the search precision; for example, in a 5-dimensional space, each space is searched 10 times, and the total search amount is 105Next, the process is carried out. If the search precision is improved by 1 time and each dimension is searched 20 times, the calculated amount is changed to the original 2564 times. An alternative approach is to search through a coarser grid, then determine a coarse parameter range, and then perform a finer search within this small range. Either grid search algorithm is strongly dependent on the accuracy of the parameter search and requires some a priori knowledge to determine the initial search offset.
Another search method is a gradient descent-based method, which determines a search direction by estimating a change in a gradient near a start point, and can quickly find an optimal parameter with a smaller amount of search. And the gradient method has the advantage that the search reliability is remarkably reduced along with the increase of the dimensionality. A grid combined gradient descent method is also often adopted, a coarse grid covering a global parameter space is used for global search, and then a gradient descent method is used for local precise search, but the search reliability of each parameter cannot be evaluated in both grid search and gradient search.
Disclosure of Invention
The invention provides a method and a system for searching reconstruction parameters in three-dimensional reconstruction of a cryoelectron microscope, which overcome the problems or at least partially solve the problems, and solve the problems that the space parameter searching speed is low, the reliability is low, and the searching reliability of each parameter cannot be evaluated in the prior art.
According to one aspect of the invention, a method for searching reconstruction parameters in three-dimensional reconstruction of a cryoelectron microscope is provided, which comprises the following steps:
constructing a parameter space, carrying out random sampling on each experimental photo in the parameter space by a Monte Carlo simulation method, and calculating the initial likelihood of the experimental photo and a given model on each sampling point;
resampling the sampling points with the initial likelihood degree larger than the set conditions to generate new sampling points, and calculating the corresponding likelihood degree;
repeating the resampling process until the mean square error of the distribution of all the sampling points is not reduced;
and taking the statistical distribution parameters of the converged sampling points as a statistical description of the reconstruction parameters of the experimental photo, and using the statistical description to reconstruct a three-dimensional electron density map of the biomacromolecule.
Preferably, the constructing the parameter space specifically includes:
constructing a translation subspace, and describing the translation subspace through two translation parameters of x and y; constructing a rotation subspace, and describing the rotation subspace by a unit quaternion q; constructing an out-of-focus quantum space, and describing the out-of-focus quantum space by a change scale coefficient zeta of an out-of-focus amount; constructing a structural state subspace, and describing the structural state subspace through an integer number mu for describing the structural state of the structural state;
and combining the translation subspace, the rotation subspace, the defocusing subspace and the structural state subspace into a parameter space { x, y, q, zeta, mu }.
Preferably, the obtaining the likelihood of each sampling point and the experimental photograph specifically includes:
and projecting the three-dimensional reference object according to the parameters of the sampling points, and calculating the likelihood between the projection and the experimental picture.
Preferably, the resampling the sample points corresponding to the projections with the likelihood greater than the set condition specifically includes:
and taking the likelihood as the weight of the sampling points, sequencing the sampling points in a high-low mode according to the weight, and resampling the N sampling points which are sequenced at the front, namely, regenerating a plurality of sampling points by taking each original sampling point as a center, and removing the sampling points which are arranged at the back and have lower weights, so that the total number of the sampling points before and after resampling is unchanged.
Preferably, the resampling the sample points with the likelihood greater than the set condition further includes:
and after sampling is finished each time, counting the distribution condition of the sampling points, and performing resampling in the next round based on the mean square error of the distribution of the sampling points.
Preferably, the step of no further reduction of the mean square error of the distribution of all the sampling points comprises:
and converging the likelihood of all sampling points, and if resampling can not make the sampling points converge to a smaller area, converging all the sampling points to be near the sampling point with the maximum likelihood.
A spatial parameter searching system in three-dimensional reconstruction of a cryoelectron microscope comprises:
the sampling module is used for carrying out random sampling on each experimental photo in a parameter space by a Monte Carlo simulation method to obtain a plurality of sampling points;
the searching module is used for projecting the three-dimensional reference object based on the parameters of each sampling point to obtain the likelihood of each projection and the experimental picture;
and the circulation module is used for sending a resampling instruction to the sampling points corresponding to the projection with the likelihood degree greater than the set condition, and sending a sampling stopping instruction when all the sampling points converge to the sampling points corresponding to the projection with the maximum likelihood degree.
And the reconstruction module is used for performing three-dimensional reconstruction according to the reconstruction parameter information reflected by the sampling points, taking the reciprocal of the mean square error of the distribution of the sampling points as the weight, and randomly selecting N or all the sampling points to participate in the three-dimensional reconstruction according to the weight.
Preferably, the system further comprises a confidence module, wherein the confidence module is used for counting the distribution condition of the sampling points in the parameter space after each round of sampling is finished, and calculating the mean square error of the distribution of the sampling points in the parameter space.
Preferably, the system further comprises a weight calculation module, wherein the weight calculation module is configured to calculate a reciprocal of a mean square error of distribution of the sampling points after all the sampling points converge to the sampling point corresponding to the projection with the maximum likelihood, and perform normalization processing, and use the reciprocal as a weight of the corresponding experimental photograph.
The invention provides a method and a system for searching reconstruction parameters in three-dimensional reconstruction of a refrigeration electron microscope, wherein the parameters are estimated by using a particle filter type key sampling algorithm in the three-dimensional reconstruction of the refrigeration electron microscope, the parameter estimation is carried out based on a random sampling method, and the confidence coefficient measurement of single parameter estimation in the three-dimensional reconstruction of the refrigeration electron microscope is realized, so that the robustness of high-dimensional parameter estimation is improved, orientation-related parameters can be more effectively searched and two-dimensional and three-dimensional classification can be carried out, the three-dimensional reconstruction resolution of some samples can be greatly improved by carrying out local search on imaging defocus parameters, meanwhile, the three-dimensional reconstruction resolution of some samples is also well adapted to defocus measurement errors caused by the thickness and the inclination of the samples during imaging, and the atomic resolution is more easily obtained.
Drawings
Fig. 1 is a schematic flow chart of a method for searching spatial parameters in three-dimensional reconstruction of a cryoelectron microscope according to an embodiment of the invention.
Detailed Description
The following detailed description of embodiments of the present invention is provided in connection with the accompanying drawings and examples. The following examples are intended to illustrate the invention but are not intended to limit the scope of the invention.
As shown in fig. 1, a method for searching reconstruction parameters in three-dimensional reconstruction of a cryoelectron microscope is shown, which includes:
constructing a parameter space, and carrying out random sampling on each experimental photo in the parameter space by a Monte Carlo simulation method to obtain the likelihood of each sampling point and the experimental photo; specifically, for a given three-dimensional reference object and an experimental photograph, the three-dimensional reference object is projected according to a set of given parameters, and the similarity between the projection and the experimental photograph is described by likelihood (likelihood). For each such parameter, its corresponding likelihood can be calculated.
And resampling the sampling points with the likelihood degree larger than the set condition until all the sampling points converge to the area with the maximum likelihood degree, and taking the distribution parameters of the sampling points as a statistical description for describing the measured reconstruction parameters. A set of globally optimal parameters is determined by searching in a parameter space, and the confidence of each parameter is calculated. By "optimal" it is meant that there is a maximum likelihood between the three-dimensional reconstructed electron density map calculated under the set of parameters and the structural information carried by the experimental photograph.
The basis of three-dimensional reconstruction of the model is the central section theorem, and the key problem in the reconstruction process is how to determine the spatial angle (orientation determination) of each particle image. Most model reconstruction and optimization algorithms are iterative methods based on projection matching (projection matching). The method comprises the steps of firstly utilizing a rough three-dimensional structure model to perform projection to obtain a reference image, comparing the reference image with an experimental particle image, updating a space orientation parameter according to a result, then constructing a new three-dimensional structure, correcting the space orientation of the experimental image, forming an iterative process, and obtaining a final three-dimensional model until convergence. In the three-dimensional reconstruction process, each orientation has translation and rotation, and a spatial orientation angle is added, so that a total of five degrees of freedom exist, and each image can be reconstructed by at least five parameters. Only after the respective orientation has been determined for each experimental picture can a three-dimensional reconstruction be performed.
The Monte Carlo method, also known as statistical simulation method, refers to a method that uses random numbers (or more commonly pseudo-random numbers) to solve many computational problems. The problem solving process of the monte carlo method can be summarized into three main steps: constructing or describing a probabilistic process; sampling from known probability distributions is achieved; various estimates are established. The idea of Particle filtering (PE) is based on the Monte Carlo method, which uses a set of particles to represent probabilities, and can be used on any form of state space model. The core idea is to express the distribution by random state particles extracted from the posterior probability distribution, and the method is a Sequential importance sampling method (Sequential inportancecamplling). Briefly, the particle filtering method is a process of approximating a probability density function by searching a group of random samples propagating in a state space, and substituting an integral operation with a sample mean value to obtain a state minimum variance distribution. The samples herein refer to particles, and any form of probability density distribution can be approximated when the number of samples N → ∞ is.
Specifically, in the present embodiment, this parameter space is divided into a plurality of subspaces, including a translation subspace, a rotation subspace, an out-of-focus subspace, and a structural state subspace. The translation subspace is described by two translation parameters, x and y; the rotational subspace is described by one unit quaternion q; the defocusing quantum space is described by a change scale coefficient zeta of the defocusing amount; the structural state subspace is a discrete parameter, and the number of the structural state is described by an integer mu. Namely, the construction of the parameter space specifically includes:
constructing a translation subspace, and describing the translation subspace through two translation parameters of x and y; constructing a rotation subspace, and describing the rotation subspace by a unit quaternion q; constructing an out-of-focus quantum space, and describing the out-of-focus quantum space by a change scale coefficient zeta of an out-of-focus amount; constructing a structural state subspace, and describing the structural state subspace through an integer number mu for describing the structural state of the structural state;
and combining the translation subspace, the rotation subspace, the defocusing subspace and the structural state subspace into a parameter space { x, y, q, zeta, mu }. On the basis, the method can more effectively search orientation-related parameters and perform two-dimensional and three-dimensional classification, and can also perform local search on imaging defocus parameters. Searching for defocus parameters can greatly improve the resolution of three-dimensional reconstruction for some samples, especially when the resolution is higher than 3 angstroms. Meanwhile, the method has good adaptability to defocusing measurement errors caused by the thickness and the inclination of the sample during imaging, and makes the atomic resolution easier to obtain.
Initializing the distribution mean square deviation sigma of the sampling points, calculating the weight of the sampling points, splitting the sampling points with high weight according to the distribution mean square deviation sigma, removing the sampling points with low weight, keeping the total number of the sampling points unchanged, specifically, in the embodiment, resampling the sampling points corresponding to the projection with the likelihood degree greater than the set condition specifically comprises:
and taking the likelihood as the weight of the sampling points, sequencing the sampling points in a high-low mode according to the weight, and resampling the N sampling points which are sequenced at the front, namely, regenerating a plurality of sampling points by taking each original sampling point as a center, and removing the sampling points which are arranged at the back and have lower weights, so that the total number of the sampling points before and after resampling is unchanged.
And repeating the resampling process until the mean square error of the distribution of all the sampling points is not reduced, so that the likelihood of all the sampling points is converged, and if the resampling can not make the sampling points converge to a smaller area, all the sampling points are converged to the vicinity of the sampling point with the maximum likelihood. The estimation of the whole parameter is carried out by a plurality of rounds of search and iteration operation. The initial search was randomly sampled in the parameter space using Monte Carlo (Monte Carlo) simulations. The likelihood of each sample point is then calculated and taken as the weight for that sample point. Each sampling point will be split according to the weight, called resampling. Each sampling point with high weight is divided into a plurality of sampling points, and the sampling points are scattered near the original sampling points. The total number of sampling points is set to be constant before and after resampling. Those samples with lower weights are removed. By repeating this sampling and resampling process, these fixed number of sampling points will gradually converge to the vicinity of the sampling point (parameter) with the maximum likelihood.
Specifically, in the iterative estimation process, the first round is an initial estimation round, and a large number of sampling points are uniformly distributed in a given parameter space. The subsequent rounds converge gradually to the vicinity of the globally optimal parameters by resampling on the basis of the first round. And when each round is finished, counting the distribution condition of the sampling points, and taking the mean square error of the counted sampling point distribution as the basis of the resampling of the next round. The statistical distribution of the sampling points in the parameter space reflects the probability density function of the parameter measurement, describing the confidence of the parameter estimation. The inverse of the mean square error of the distribution of sample points is calculated in the final round and, after normalization, taken as the weight of this picture for adjusting its contribution in the three-dimensional reconstruction.
The weight adjustment in three-dimensional reconstruction is embodied in two aspects. One is to directly multiply the pixel values of the experimental photograph by the weight. The other is that each photo is used for a plurality of times in three-dimensional reconstruction, but the corresponding parameters are different in each use. These parameters are randomly obtained and have the same distribution as the sampling points of the final round.
The method realizes the confidence measurement of single parameter estimation in the three-dimensional reconstruction of the cryoelectron microscope, thereby improving the robustness of high-dimensional parameter estimation. The confidence of the parameter evaluation is further used as a weighting factor for adjusting the contribution of the current photograph to the final three-dimensional reconstruction. The function greatly improves the tolerance of the method to bad photos and reduces the difficulty of image screening. The method is beneficial to the implementation of automatic three-dimensional reconstruction no matter the defocusing amount parameter is refined or the bad picture is highly tolerant, and more importantly, the method provides guarantee for future industrialized large-scale high-throughput biomacromolecule structure determination.
The embodiment further provides a system for searching the reconstruction parameters in the three-dimensional reconstruction of the cryoelectron microscope, which comprises:
the sampling module is used for carrying out random sampling on each experimental photo in a parameter space by a Monte Carlo simulation method to obtain a plurality of sampling points;
the searching module is used for projecting the three-dimensional reference object based on the parameters of each sampling point to obtain the likelihood of each projection and the experimental picture;
and the circulation module is used for sending a resampling instruction to the sampling points corresponding to the projection with the likelihood degree greater than the set condition, and sending a sampling stopping instruction when all the sampling points converge to the sampling points corresponding to the projection with the maximum likelihood degree.
And the reconstruction module is used for performing three-dimensional reconstruction according to the reconstruction parameter information reflected by the sampling points, taking the reciprocal of the mean square error of the distribution of the sampling points as the weight, and randomly selecting N or all the sampling points to participate in the three-dimensional reconstruction according to the weight.
In this embodiment, the system further includes a confidence module, where the confidence module is configured to count distribution of the sampling points in the parameter space after each round of sampling is finished, and calculate a mean square error of distribution of the sampling points in the parameter space.
In this embodiment, the method further includes a weight calculation module, where the weight calculation module is configured to calculate a reciprocal of a mean square error of distribution of the sampling points after all the sampling points converge to the sampling point corresponding to the projection with the maximum likelihood, and perform normalization processing, where the reciprocal is used as a weight of the corresponding experimental photograph.
In summary, the invention provides a method and a system for searching spatial parameters in three-dimensional reconstruction of a cryo-electron microscope, wherein a particle filter-type key sampling algorithm is used for estimating parameters in the three-dimensional reconstruction of the cryo-electron microscope, and the parameter estimation is performed based on a random sampling method, so that the confidence measurement of single parameter estimation in the three-dimensional reconstruction of the cryo-electron microscope is realized, the robustness of high-dimensional parameter estimation is improved, orientation-related parameters can be more effectively searched and two-dimensional and three-dimensional classification can be performed, local search can be performed on imaging defocus parameters, the three-dimensional reconstruction resolution of some samples can be greatly improved by searching the defocus parameters, meanwhile, the method and the system have good adaptability on defocus measurement errors caused by the thickness and inclination of the samples during imaging, and the atomic resolution is more easily obtained.
Finally, the method of the present invention is only a preferred embodiment and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A method for searching reconstruction parameters in three-dimensional reconstruction of a cryoelectron microscope is characterized by comprising the following steps:
constructing a parameter space, and randomly sampling each experimental photo in the parameter space by a Monte Carlo simulation method to obtain the initial likelihood of the experimental photo on each sampling point and a given model;
resampling the sampling points with the initial likelihood degree larger than the set conditions to generate new sampling points, and calculating the corresponding likelihood degree;
repeating the resampling process until the mean square error of the distribution of all the sampling points is not reduced;
taking the statistical distribution parameter of the converged sampling points as a statistical description of the reconstruction parameter of the experimental photo and reconstructing a three-dimensional electron density map of the biomacromolecule, wherein the statistical distribution parameter is the statistical distribution condition of the sampling points in a parameter space;
the constructing of the parameter space specifically includes:
constructing a translation subspace, and describing the translation subspace through two translation parameters of x and y; constructing a rotation subspace, and describing the rotation subspace by a unit quaternion q; constructing an out-of-focus quantum space, and describing the out-of-focus quantum space by a change scale coefficient zeta of an out-of-focus amount; constructing a structural state subspace, and describing the structural state subspace through an integer number mu for describing the structural state of the structural state;
combining the translation subspace, the rotation subspace, the defocusing quantum space and the structural state subspace into a parameter space { x, y, q, zeta, mu };
the statistical distribution parameter of the converged sampling points is used as a statistical description of the reconstruction parameter of the experimental photo, and the method further comprises the following steps:
calculating the reciprocal of the mean square error of the distribution of the sampling points, carrying out normalization processing, and taking the reciprocal as the weight of the corresponding experiment photo; and randomly selecting N or all the sampling points, and participating in three-dimensional reconstruction according to the same weight.
2. The method according to claim 1, wherein obtaining the likelihood of the experimental photograph with the given model at each sample point comprises:
and projecting the given three-dimensional reference object according to the parameters of the sampling points, and calculating the likelihood between the projection and the experimental picture.
3. The method of claim 1, wherein resampling sample points with initial likelihood greater than a set condition specifically comprises:
and taking the initial likelihood as the weight of the sampling points, sequencing the sampling points according to the weight, taking N sampling points which are sequenced at the front as original sampling points to perform resampling, and taking each original sampling point as a center to obtain a plurality of sampling points again, wherein the total number of the sampling points after resampling is the same as the total number of the sampling points before resampling.
4. The method of claim 1, wherein resampling the sample points with initial likelihood greater than the set condition further comprises:
and after sampling is finished each time, counting the distribution condition of the sampling points, and performing resampling in the next round based on the mean square error of the distribution of the sampling points.
5. The method of claim 1, wherein the step of reducing the mean square error of the distribution of all the sampling points comprises:
and converging the likelihood of all sampling points, and if resampling can not make the sampling points converge to a smaller area, converging all the sampling points to be near the sampling point with the maximum likelihood.
6. A reconstruction parameter search system in three-dimensional reconstruction of a cryoelectron microscope is characterized by comprising:
the construction module is used for constructing a parameter space, carrying out random sampling on each experimental photo in the parameter space by a Monte Carlo simulation method, and obtaining the initial likelihood of the experimental photo on each sampling point and a given model;
the sampling module is used for resampling the sampling points with the initial likelihood degree larger than the set conditions, generating new sampling points and calculating the corresponding likelihood degree;
a cycle module for repeating the resampling process until the mean square error of distribution of all sampling points is not reduced;
the reconstruction module is used for taking the statistical distribution parameters of the converged sampling points as a statistical description of the reconstruction parameters of the experimental picture and reconstructing a three-dimensional electron density map of the biomacromolecule, wherein the statistical distribution parameters are the statistical distribution conditions of the sampling points in a parameter space;
the constructing of the parameter space specifically includes:
constructing a translation subspace, and describing the translation subspace through two translation parameters of x and y; constructing a rotation subspace, and describing the rotation subspace by a unit quaternion q; constructing an out-of-focus quantum space, and describing the out-of-focus quantum space by a change scale coefficient zeta of an out-of-focus amount; constructing a structural state subspace, and describing the structural state subspace through an integer number mu for describing the structural state of the structural state;
combining the translation subspace, the rotation subspace, the defocusing quantum space and the structural state subspace into a parameter space { x, y, q, zeta, mu };
the statistical distribution parameter of the converged sampling points is used as a statistical description of the reconstruction parameter of the experimental photo, and the method further comprises the following steps:
calculating the reciprocal of the mean square error of the distribution of the sampling points, carrying out normalization processing, and taking the reciprocal as the weight of the corresponding experiment photo; and randomly selecting N or all the sampling points, and participating in three-dimensional reconstruction according to the same weight.
7. The system of claim 6, wherein obtaining the likelihood of the experimental photograph from the given model at each sample point comprises:
and projecting the given three-dimensional reference object according to the parameters of the sampling points, and calculating the likelihood between the projection and the experimental picture.
8. The system of claim 6, wherein resampling the sample points with initial likelihood greater than a set condition specifically comprises:
and taking the initial likelihood as the weight of the sampling points, sequencing the sampling points according to the weight, taking N sampling points which are sequenced at the front as original sampling points to perform resampling, and taking each original sampling point as a center to obtain a plurality of sampling points again, wherein the total number of the sampling points after resampling is the same as the total number of the sampling points before resampling.
9. The system of claim 6, wherein resampling the sample points with initial likelihood greater than the set condition further comprises:
and after sampling is finished each time, counting the distribution condition of the sampling points, and performing resampling in the next round based on the mean square error of the distribution of the sampling points.
10. The system of claim 6, wherein the step of until the mean square error of the distribution of all the sampling points does not decrease further comprises:
and converging the likelihood of all sampling points, and if resampling can not make the sampling points converge to a smaller area, converging all the sampling points to be near the sampling point with the maximum likelihood.
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 CN108510436A (en) | 2018-09-07 |
| CN108510436B true 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) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| 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 |
| 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 |
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 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108510436A (en) | 2018-09-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108510436B (en) | Method and system for searching reconstruction parameters in three-dimensional reconstruction of cryoelectron microscope | |
| Bendory et al. | Single-particle cryo-electron microscopy: Mathematical theory, computational challenges, and opportunities | |
| Sigworth | Principles of cryo-EM single-particle image processing | |
| Penczek | Resolution measures in molecular electron microscopy | |
| Münch et al. | FIB‐Nanotomography of Particulate Systems—Part II: Particle Recognition and Effect of Boundary Truncation | |
| EP3465552A1 (en) | Systems and methods incorporating a neural network and a forward physical model for semiconductor applications | |
| CA3110199A1 (en) | Systems and methods for enhanced imaging and analysis | |
| KR20200032651A (en) | Apparatus for three dimension image reconstruction and method thereof | |
| CN110032761B (en) | Classification method for single-particle imaging data of frozen electron microscope | |
| Jonić | Cryo-electron microscopy analysis of structurally heterogeneous macromolecular complexes | |
| CN106203511B (en) | A kind of image similar block appraisal procedure | |
| Lück et al. | Three‐dimensional analysis of intermediate filament networks using SEM tomography | |
| Brubaker et al. | Building proteins in a day: Efficient 3D molecular reconstruction | |
| Rangan et al. | Deep reconstructing generative networks for visualizing dynamic biomolecules inside cells | |
| CN112837415A (en) | Three-dimensional structure reconstruction method of protein cryoelectron microscope projection image | |
| Ge et al. | A point cloud registration method combining enhanced particle swarm optimization and iterative closest point method | |
| Bobkov et al. | Software Platform for European XFEL: Towards Online Experimental Data Analysis | |
| TW202319735A (en) | Defect examination on a semiconductor specimen | |
| Zhao et al. | Three-dimensional microstructural image synthesis from 2d backscattered electron image of cement paste | |
| Lyckegaard | Development of tomographic reconstruction methods in materials science with focus on advanced scanning methods | |
| Lu et al. | EMDiffuse: a diffusion-based deep learning method augmenting ultrastructural imaging and volume electron microscopy | |
| Nakano et al. | Molecular size dependence on achievable resolution from XFEL single-particle 3D reconstruction | |
| Nashed et al. | End-to-end simultaneous learning of single-particle orientation and 3d map reconstruction from cryo-electron microscopy data | |
| Still et al. | Alpha Shape Analysis: Extracting Composition, Surface Area, and Volume Post Clustering | |
| Pichkur et al. | Towards on-the-fly cryo-electron microscopy data processing by high performance data analysis |
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 |