JP4171919B2 - Structure estimation system, structure estimation method and program - Google Patents

Description

  The present invention relates to a system, method, and program for estimating a three-dimensional structure of a molecule.

  As a technique for analyzing the three-dimensional structure of a molecule, a sample obtained by crystallizing a biopolymer such as a protein to be analyzed has been prepared, and this sample is analyzed and evaluated by an X-ray crystal structure analysis method and an NMR method. There is a way. However, since this method requires crystallization of the sample, it could not be applied to analysis of a sample that is difficult to crystallize membrane protein or the like.

Therefore, as a method that does not require crystallization of a sample, a single particle structure analysis method has been proposed in which a protein structure analysis is performed using a cryogenic transmission electron microscope (Cryo-TEM) (Non-patent Document 1). The single particle structure analysis method cuts out images of individual globular protein samples from TEM images obtained from particulate protein samples cooled to cryogenic temperatures in a transmission electron microscope (TEM), and particles are extracted from these sample images. Is a method for estimating the three-dimensional structure of a protein sample. According to this method, the noise of the sample image can be reduced by averaging the sample images.
Sato, and others, “Protein structure analysis without using crystals by single particle analysis: taking the structure of voltage-dependent Na + channel as an example”, Electron Microscope, 2002, Vol. 37, No. 1, p. 40-44

  However, when the molecular structure analysis is performed using the sample image data as in the single particle structure analysis method, the amount of data necessary for the structure analysis and the amount of data processing are enormous. For this reason, the load on the user's terminal requesting the analysis and the terminal performing the data processing is large, and the analysis takes a long time.

  The present invention has been made in view of the above circumstances, and provides a technique for efficiently performing data processing in molecular structure analysis using image data.

  The present inventor has intensively studied to improve the efficiency and speed of image processing in molecular structure analysis using image data. As a result, by applying predetermined distributed processing to data processing using a plurality of measurement images and a plurality of reference images, the load generated during data processing can be reduced and rapid analysis processing can be performed. The headline, the present invention has been reached.

According to the present invention,
A system for estimating the three-dimensional structure of a molecule from data of a plurality of images,
A user terminal that accepts a user request and requests a structural analysis based on a plurality of measurement images of the molecule;
A data processing system comprising a plurality of data processing terminals;
A scheduler that accepts a request from the user terminal and requests data processing from the data processing system;
Are connected via a network,
The data processing system is
A plurality of reference data storage units to which data of a plurality of reference images are distributed, and a plurality of data processing units,
Each of the data processing units translates the measurement image data or the translation image data obtained by translating the measurement image for each of the plurality of measurement images in response to the data processing request from the scheduler. And a reference data set composed of data of the reference image assigned to itself and data of a group of rotationally moving images obtained by rotating and moving the reference image, and assigned to the measurement image. There line processing for evaluating the similarity between the reference image,
The scheduler
A measurement data classifying unit that associates a plurality of the measurement images with any one of the plurality of reference images based on the result of the processing for evaluating the similarity;
The averaged image data obtained by averaging the plurality of measurement images associated with the same reference image in the measurement data classifying unit is acquired for the plurality of reference images, and the plurality of averaged images acquired A structure estimation unit for estimating the three-dimensional structure of the molecule based on
A structure estimation system is provided.

Moreover, according to the present invention,
A method for estimating the three-dimensional structure of a molecule from data of a plurality of images,
A user terminal accepting a user request and requesting a structural analysis of the molecule based on a plurality of measurement images;
A scheduler accepting a request from the user terminal and requesting data processing from a data processing system including a plurality of data processing terminals;
A step wherein the scheduler, the multiple's Reference data storage unit included in the data processing system, for distributing the data of a plurality of reference images,
A step wherein the scheduler is to allocate the data processing plurality of data processing units included in the system, the process of evaluating the similarity between a plurality of measurement images and the plurality of reference images,
Each of the plurality of data processing units is assigned to the data of the measurement image or the data of the translation image obtained by translating the measurement image in response to the data processing request from the scheduler . by comparing the reference data set composed of a data group of rotational movement image composed by rotating moving data and the reference image of the reference image, similar to the reference image allocated to the measurement image Performing the process of evaluating the degree;
The scheduler associates each of the plurality of measurement images with one of the plurality of reference images based on the result of the process of evaluating the similarity,
The scheduler acquires a plurality of averaged image data obtained by averaging the plurality of measurement images associated with any of the plurality of reference images, and based on the averaged image data, Estimating the conformation of the molecule;
A structure estimation method is provided.

  In the present invention, data of a plurality of reference images is distributed to a plurality of reference data storage units. A plurality of data processing units share and perform data processing related to the evaluation of the similarity between the measurement image and the plurality of reference images. Since enormous data processing required when using a plurality of measurement images and a plurality of reference images is distributed to a plurality of data processing units, an increase in local load in the data processing system can be reduced. For this reason, data processing can be performed efficiently and the processing speed can be increased. Moreover, since it is not necessary to perform data processing in the user terminal, the load on the user terminal can be reduced.

  In the structure estimation system of the present invention, a plurality of the averaged image data are redistributed as updated reference images to the plurality of reference data storage units, and the updated reference images are allocated to the plurality of data processing units. The data processing unit performs a process of evaluating the similarity between the measurement image and the assigned update reference image, and the measurement data classifying unit performs a plurality of processes based on the result of the process of evaluating the similarity. A plurality of the measurement images are associated with any of the updated reference images, and the structure estimation unit averages the plurality of measurement images associated with the same updated reference image in the measurement data classification unit. The averaged image data obtained by acquiring the plurality of the updated reference images, the plurality of acquired It can be configured to estimate the three-dimensional structure of the molecule based upon the disproportionation image. In this configuration, a plurality of updated reference images are used as reference images when estimating the three-dimensional structure of molecules. For this reason, the precision of the reference image used for structure estimation can be further improved.

  In the structure estimation system of the present invention, the scheduler selects, based on operating states of the plurality of data processing units, data for selecting a plurality of data processing units to which the plurality of reference images are assigned for the processing for evaluating the degree of similarity. You may include a process part selection part and the reference distribution information storage part which memorize | stores the information regarding the distribution condition of the said some reference data set. In the structure estimation system of the present invention, the data processing unit selection unit may further select a plurality of the data processing units that acquire the reference data set. By doing so, the data processing can proceed more efficiently.

  In the structure estimation system of the present invention, a plurality of the data processing units select a reference data set acquisition unit for acquiring the reference data set, and a reference data set for selecting the reference data set acquisition method in the reference data set acquisition unit And an acquisition method selection unit. In the structure estimation system of the present invention, the plurality of data processing units include a reference data set acquisition unit that acquires the reference data set, and the scheduler acquires the reference data set in the reference data set acquisition unit. A reference data set acquisition method selection unit that selects a method may be included. In this configuration, since the acquisition of the reference data set is distributed to a plurality of data processing units, the reference data can be acquired more efficiently.

  In the present invention, the reference data set may include data of the reference image and data of a rotationally moving image group obtained by rotating the reference image from 0 degrees to 360 degrees. Thereby, it is possible to more reliably evaluate the similarity between the reference image and the measurement image. In the present invention, the reference data set may include data of the reference image and data of a rotationally moving image group obtained by rotating the reference image from 0 degrees to 90 degrees. In this way, it is only necessary to create a four-fold symmetric image group from among the rotationally moving image group, and the number of rotationally moving images to be created in advance is reduced as necessary when evaluating the degree of similarity. Rotation image data up to 360 degrees can be easily acquired by inversion. For this reason, the load on the system can be further reduced and the processing can be performed at higher speed.

  In the present invention, the molecule may be a biopolymer. Furthermore, in this configuration, the biopolymer may be a membrane protein. In this way, since the three-dimensional structure can be estimated without crystallizing the membrane protein, the three-dimensional structure of a membrane protein that is generally difficult to crystallize can be reliably estimated.

  It should be noted that any combination of the above-described constituent elements and a conversion of the expression of the present invention between a method, an apparatus, a system, a recording medium, a computer program, etc. are also effective as an aspect of the present invention.

For example, according to the present invention,
A system for estimating a three-dimensional structure of a molecule from data of a plurality of images,
A user terminal that accepts a user request and requests a structural analysis based on a plurality of measurement images of the molecule;
A data processing system comprising a plurality of data processing terminals;
A scheduler that accepts a request from the user terminal and requests data processing from the data processing system;
Are connected via a network,
The data processing system is
A plurality of reference data storage units to which data of a plurality of reference images are distributed, and a plurality of data processing units,
Each of the data processing units translates the measurement image data or the translation image data obtained by translating the measurement image for each of the plurality of measurement images in response to the data processing request from the scheduler. And a reference data set composed of data of the reference image assigned to itself and data of a group of rotationally moving images obtained by rotating and moving the reference image, and assigned to the measurement image. There line processing for evaluating the similarity between the reference image,
The scheduler
A measurement data classifying unit that associates a plurality of the measurement images with any one of the plurality of reference images based on the result of the processing for evaluating the similarity;
The averaged image data obtained by averaging the plurality of measurement images associated with the same reference image in the measurement data classification unit is acquired for the plurality of reference images, and the plurality of the averaged acquired A structure estimation unit that estimates the three-dimensional structure of the molecule based on an image;
There is provided a program that functions as a structure estimation system including.

  ADVANTAGE OF THE INVENTION According to this invention, the technique which performs a data processing efficiently in the structure analysis of the molecule | numerator using image data is implement | achieved.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In all the drawings, common constituent elements are denoted by the same reference numerals, and description thereof is omitted as appropriate. In the following embodiments, a case where a three-dimensional structure of a biopolymer is analyzed using a single particle analysis method will be described as an example. However, the analysis target of the present invention is not limited to a biopolymer, and synthesis is performed. Polymers and various low-molecular compounds can also be analyzed. In addition, the present invention can be applied to techniques other than the single particle analysis method as long as the three-dimensional structure analysis of molecules is performed by image processing.

  First, in order to deepen the understanding of the present invention, an outline of a conventional single particle analysis method will be described with reference to FIG. 7 as an example of a method for analyzing a three-dimensional structure of a molecule. FIG. 7 is a diagram for explaining the outline of the single particle analysis method disclosed in Non-Patent Document 1 described above in the background art section. As described in this document, the single particle analysis method is a protein structure analysis method that does not use crystals. In the single particle analysis method, structural analysis is performed using an electron micrograph of a protein.

  An electron micrograph of a protein is a noisy image with limited resolution due to sample damage due to electron beam irradiation. When proteins are susceptible to electron beam damage, the resolution of the image is limited due to the poor signal-to-noise ratio (S / N ratio), so the original resolution of the electron microscope cannot be achieved. . For this reason, not only high-resolution information cannot be obtained from one image, but also the reconstruction of the three-dimensional structure is not easy.

  In order to overcome this problem, in single particle analysis, a method that should be referred to as crystallization using a computer, which selects particle images in the same direction and averages them to improve the S / N ratio and increase the resolution. Used (in FIG. 7). Since the unstained electron microscope image by the freezing method is thought to give a projected image reflecting the intramolecular density, create a sinogram projected in one dimension at various angles from the average image with improved S / N ratio, By comparing with each other, each other's three-dimensional angle is estimated (lower left of FIG. 7).

  In other words, here, components having a high degree of coincidence are searched for using a sinogram projected in a one-dimensional manner by changing the angle with respect to the entire circumference of 360 degrees. The images with similar particle projection directions are overlapped and averaged. The mutual values are obtained for three or more averaged images, and the relative azimuth angle of the particles in each averaged image is estimated. Based on this, a three-dimensional structure with less noise is reconstructed by combining the average projected images three-dimensionally (lower right of FIG. 7). Since the noise is random and does not come to the same position, the S / N ratio can be further improved by averaging many images.

  Further, the obtained plurality of averaged images are used again as a reference image for alignment of the original image, and a newly improved averaged image is generated. By repeating such a cycle a predetermined number of times, the S / N ratio is further increased.

  The single particle analysis method is performed by the above procedure. However, in this method, it is necessary to create a parallel movement image and a rotation movement image of the target particle at the time of analysis. In addition, alignment between these images and the reference image is necessary. If all these processes are performed by a single terminal, the amount of data processing is too large, the load on the terminal is large, and analysis takes a long time. In addition, this problem becomes more prominent when the averaged image creation cycle is repeated a plurality of times.

  Therefore, in the present invention, instead of creating a rotationally moving image of a measurement image, a rotationally moving image of a reference image (reference image) is created in advance for a plurality of reference images, thereby searching for a component having high similarity. In addition, it is unnecessary to create a rotational movement image of the target particles. A data processing system having a plurality of data processing units is used. Then, distributed processing is performed by a plurality of predetermined data processing units for each process of obtaining a rotationally moving image of the reference image and evaluating the similarity. As a result, the local load is reduced, and a huge amount of data processing can be performed quickly and efficiently. A specific configuration will be described below.

  FIG. 1 is a diagram illustrating a basic configuration of a structure estimation system according to an embodiment of the present invention. A structure estimation system 100 illustrated in FIG. 1 is a system that estimates a three-dimensional structure such as a biopolymer, and includes a user terminal 101, a scheduler 103, and a data processing system 110. Hereinafter, the case where the three-dimensional structure of a biopolymer is estimated will be described as an example.

  The user terminal 101 receives a user service use request and requests a structural analysis of the biopolymer. The scheduler 103 receives a request from the user terminal 101 and requests data processing from the data processing terminal group 105. Further, the scheduler 103 estimates the structure of the biopolymer based on the information obtained by the data processing and presents it to the user terminal.

  The data processing system 110 performs data processing using the measurement image data of the biopolymer. The data processing system 110 includes a plurality of data processing terminals and includes a plurality of reference data storage units 113 and a plurality of data processing units 111. The plurality of reference data storage units 111 and the plurality of data processing units 113 are respectively provided in a predetermined plurality of data processing terminals 105.

  FIG. 1 shows an example in which the data processing system 110 includes five terminals, that is, a data processing terminal 105a to a data processing terminal 105b. The data processing terminal 105a, the data processing terminal 105b, and the data processing terminal 105d include a data processing unit 113 and a reference data storage unit 111. The data processing terminal 105c and the data processing terminal 105e include only the data processing unit 113. Although not shown in FIG. 1, the structure estimation system 100 may include a terminal that does not include the second storage unit 109 but includes the reference data storage unit 111.

  The structure estimation system 100 is a system that estimates a three-dimensional structure of a molecule from a plurality of images. In the structure estimation system 100, the three-dimensional structure of the sample is reconstructed using a plurality of measurement images having different observation directions. When the three-dimensional structure is reconstructed, a plurality of reference images having a smaller number than the measurement image is used. The similarity between the measurement image and the reference image is evaluated by matching the plurality of measurement images with the reference image. The measurement image is classified for each reference image having the highest similarity. Then, a plurality of measurement images classified into the same reference image are averaged to obtain an averaged image. The obtained averaged image is updated as a new reference image. By repeating the update process, the accuracy of the reference image is improved. The three-dimensional structure of the sample is accurately reconstructed using a plurality of reference images updated a predetermined number of times.

In the structure estimation system 100, among the processes described above,
The processing related to (i) creation of a reference data set and (ii) evaluation of the degree of similarity between the measurement image and the reference data set is distributed to a plurality of data processing units 113.

  In (i) above, the reference data set is data of one reference set. One reference set is composed of one reference image and a group of images obtained by rotating the reference image at a predetermined pitch (rotation-moving image group). One reference data set includes data of one reference image and data of a rotationally moving image group of the reference image. The plurality of reference data sets are distributed to the plurality of reference data storage units 111.

For example, in FIG. 1, reference image data ref. a to ref. i reference data sets ref. seta-ref. seti is created. For example, when the rotational angle pitch of a rotationally moving image is 1 degree and a rotationally moving image up to 90 degrees is created,
ref. seta: ref. a0 (= ref.a), ref. a1, ref. a2,... ref. a90,
ref. setb: ref. b0 (= ref.b), ref. b1, ref. b2, ..., ref. b90,
...
ref. seti: ref. i0 (= ref.i), ref. i1, ref. i2, ..., ref. i90,
Thus, each reference set is composed of data of 91 images. The data processing terminal 105a, the data processing terminal 105b, and the data processing terminal 105d have ref. seta-c, ref. setd-f, and ref. setg to i are distributed. The reference data set is created by distributed processing in the plurality of data processing units 113. Thereby, the local load of the data processing system 110 can be reduced, and a reference data set corresponding to each of a plurality of reference images can be efficiently created.

  In the above (ii), the plurality of measurement images are a group of images obtained by viewing one sample from a plurality of directions. The degree of similarity with each reference image is evaluated for each of the data of the plurality of measurement images. The evaluation of the similarity is performed by comparing the measurement image data with the reference data set. At this time, by assigning a plurality of reference images to a plurality of data processing units 113, it is possible to efficiently evaluate the similarity. In FIG. 1, each of the data processing terminal 105a, the data processing terminal 105b, the data processing terminal 105c, the data processing terminal 105d, and the data processing terminal 105e is set to ref. ab, ref. cd, ref. ef, ref. g to h, and ref. i is assigned as the person in charge. Each data processing terminal performs matching with the assigned reference set for each of the plurality of measurement images.

  FIG. 2 is a functional block diagram showing the configuration of the structure estimation system 100 shown in FIG. FIG. 3 is a functional block diagram showing detailed structures of the scheduler 103 and the data processing terminal 105 of the structure estimation system 100. Hereinafter, the configuration of the structure estimation system 100 will be described in more detail with reference to FIGS. 2 and 3.

  The structure estimation system 100 is a system in which a user terminal 101, a scheduler 103, and a data processing system 110 are connected by a network 115.

  The user terminal 101 includes a first storage unit 107 and a first transmission / reception unit 117. The first storage unit 107 stores measurement data of a cryo-TEM image of the sample.

  The data processing terminal 105 is a terminal that constitutes a data processing terminal group. FIG. 2 illustrates the case where the data processing system 110 includes three data processing terminals 105 including both the reference data storage unit 111 and the data processing unit 113. However, a plurality of reference data storage units 111 and a plurality of data are included. If the data processing system 110 includes the processing unit 113, the number of data processing terminals 105 is not particularly limited. The data processing terminal 105 includes a second transmission / reception unit 119, a data processing unit 113, and a reference data storage unit 111. The reference data storage unit 111 is provided in the second storage unit 109 (FIG. 3).

  The second transmission / reception unit 119 transmits / receives measurement image data, reference data, data processing results, and the like via the network 115.

  The data processing unit 113 rotates, for each of the plurality of measurement images, the measurement image data or translation image data obtained by translating the measurement image, the assigned reference image data, and the reference image. And a reference data set composed of data of a group of rotationally moving images, and a process of evaluating the similarity between the measured image and the assigned reference image. In addition, when the reference image is updated, the updated reference image is assigned to the data processing unit 113. Then, the data processing unit 113 performs processing for evaluating the similarity between the measurement image and the assigned update reference image.

  The data processing unit 113 includes a translation image acquisition unit 141, a reference data set acquisition method selection unit 143, a reference data set acquisition unit 145, a similarity parameter assignment unit 147, and an averaging processing unit 149.

  The translation image acquisition unit 141 acquires image data obtained by translating the measurement image data in a predetermined direction in the plane. The translation image acquisition unit 141 refers to the calculation formula stored in the translation image information storage unit 135, for example, and translates the measurement image in a predetermined pitch in each of the X axis direction and the Y axis direction. Data of the group of moving images is created and stored in the parallel moving image information storage unit 135 in association with the identifier of the measurement image.

  The reference data set acquisition method selection unit 143 selects a reference data set acquisition method in the reference data set acquisition unit 145. For example, it communicates with another data processing terminal 105 via the network 115, acquires the reference data set stored in the second storage unit 109 according to the communication time, or rotates and moves in its own reference data set acquisition unit 145 Choose between creating image data and a reference data set.

  The reference data set acquisition unit 145 acquires a reference data set. For example, the data stored in the reference data set storage unit 139 in the data processing terminal 105 constituting the data processing terminal group is acquired. Alternatively, based on the reference image data distributed from the scheduler 103, the reference data set acquisition unit 145 creates data of the rotationally moving image group, and the generated rotationally moving image group data and the reference image data before the rotation Is associated with the identifier of the reference image as a reference data set and stored in the reference data set storage unit 139.

  The similarity parameter assigning unit 147 aligns the measurement image data and the translation image data with the reference data set. Also, by comparing these, a similarity parameter indicating the similarity between the reference image data and the measurement image data is calculated. When calculating the similarity parameter, the reference image data and the measurement image data may be vectorized using a known method. At this time, the similarity parameter is obtained by calculating the inner product of the vectorized reference image data and the measurement image data. Also, a parameter reflecting the inner product may be calculated and used as a similarity parameter. When calculating the inner product, the n × n data matrix constituting each image may be diagonalized. When the S / N ratio of the measurement image data is low, filtering by a known method may be performed prior to the calculation of the similarity parameter. The obtained similarity parameter is stored in the similarity information storage unit 151 in association with the identifier of the measurement image.

  For each reference image, the averaging processing unit 149 performs averaging processing on data of a plurality of measurement images classified as having a high similarity (optimum similarity) with respect to the reference image. Data obtained by the averaging is stored in the average information storage unit 153 and sent to the scheduler 103 via the second transmission / reception unit 119.

  An example of the averaging processing method in the averaging processing unit 149 is a method of obtaining an arithmetic average of data of a plurality of measurement images for each dot of an n × n matrix. The averaging process may include a noise removal process. As the noise removal processing, for example, noise averaging that removes noise by approximating it to white noise can be cited.

  The second storage unit 109 includes a measurement data storage unit 131, a reference data storage unit 111, a similarity information storage unit 151, and an averaged information storage unit 153.

  The measurement data storage unit 131 includes a measurement image data storage unit 133 and a parallel movement image information storage unit 135.

  The measurement image data storage unit 133 stores measurement image data. The measurement image data is acquired based on the distribution information of the measurement image data presented from the scheduler 103. The translation image information storage unit 135 stores a conversion formula for converting measurement image data into translation image data. Usually, data of a group of a plurality of translation images is created from one measurement data. In addition, data of the created parallel movement image may be stored in the parallel movement image information storage unit 135.

  The reference data storage unit 111 includes a reference image data storage unit 137 and a reference data set storage unit 139.

  In the plurality of reference image data storage units 137 of the data processing system 110, data of a plurality of reference images is distributed. Further, when the reference image is updated, the data of the plurality of averaged images is redistributed as the updated reference image. Details of the reference data update process will be described later.

  In the reference data set storage unit 139, reference set data acquired from the distributed reference image, that is, a reference data set is stored in association with the identifier of the reference image data. The reference set is a group of images obtained by rotating the reference image by 0 ° to 360 °, for example, at a pitch of 1 °. Note that the image rotated by 0 degrees corresponds to the original reference image. The rotation angle of the reference image is preferably in the range of at least 0 degrees to 90 degrees. At this time, the reference data set includes reference image data and data of a rotationally moving image group obtained by rotating the reference image from 0 degrees to 90 degrees. If rotation data of 0 to 90 degrees is created in advance, the amount of data can be significantly reduced as compared with the case where all rotated images up to 360 degrees are stored. In addition, if rotation data of 0 to 90 degrees corresponding to four-fold symmetry is created in advance, the similarity between the measurement image and the reference image is evaluated by sign inversion in the X-axis direction or the Y-axis direction. Therefore, it is possible to easily acquire the data of the 0 degree to 360 degree rotated image necessary for the post-processing. The pitch angle of the rotated image is, for example, 5 degrees or less, and more specifically, 1 degree. In this way, it is possible to improve the evaluation accuracy of the similarity between the measurement data and the reference data set.

  The similarity information storage unit 151 stores a calculation formula for calculating a parameter indicating the similarity between the reference image data and the measurement degree data.

  The averaged information storage unit 153 stores, for each reference image, a calculation formula used when averaging data of measurement images classified as having the highest similarity to the reference image.

  The scheduler 103 includes a third transmission / reception unit 121, a calculation unit 123, and a third storage unit 129.

  The calculation unit 123 includes a data processing unit selection unit 155, a measurement data classification unit 157, and a structure estimation unit 125.

  The data processing unit selection unit 155 selects a plurality of data processing units 113 to which a plurality of reference images are assigned for the process of evaluating the similarity based on the operating states of the plurality of data processing units 113. In addition, the data processing unit selection unit 155 further selects a plurality of data processing units 113 that acquire the reference data set.

That is, the data processing unit selection unit 155 selects the data processing unit 113 that is in charge of the data processing related to (i) creation of the reference data set and (ii) evaluation of the similarity between the measurement image and the reference data set. . At this time, for example, the data processing unit selection unit 155 detects the load of each data processing unit 113 and selects the data processing unit 113 having a light load. Further, CPU occupancy data of the data processing terminal 105 may be acquired, or operating state information may be acquired. In addition, the data processing unit selection unit 155 may acquire data of the time required for communication with the data processing unit 113 and select the data processing unit 113 of the data processing terminal 105 capable of transmitting data in a short time. it can.

  The measurement data classification unit 157 associates a plurality of measurement images with any one of the plurality of reference images based on the result of the process for evaluating the similarity. Specifically, the measurement data classifying unit 157 acquires the measurement image data to which the similarity parameter is assigned, selects the reference image with the highest degree of coincidence for each of the plurality of measurement images, and selects the reference image Classification data in which the identifier and the identifier of the measurement image are associated with each other is created and stored in the calculation information storage unit 163. Further, when updating the reference, a plurality of measurement images are associated with one of the plurality of updated reference images based on the result of the process of evaluating the similarity.

  The structure estimation unit 125 reconstructs the three-dimensional structure of the membrane protein to be analyzed based on the updated reference images created by the averaging processing unit 149. At this time, the structure estimation unit 125 acquires, for a plurality of reference images, averaged image data obtained by averaging a plurality of measurement images associated with the same reference image in the measurement data classification unit 157. The three-dimensional structure of the molecule is estimated based on the plurality of averaged images. In addition, the structure estimation unit 125 may create stereoscopic image data based on a plurality of plane image data with reference to an arithmetic expression stored in the arithmetic information storage unit 163. Furthermore, when updating the reference, the structure estimation unit 125 sets a plurality of averaged image data obtained by averaging a plurality of measurement images associated with the same updated reference image in the measurement data classification unit 157. The updated reference image is acquired, and the three-dimensional structure of the molecule is estimated based on the acquired averaged images.

  The third storage unit 129 includes a reference information storage unit 159, a measurement data arrangement information storage unit 161, and a calculation information storage unit 163.

  The reference information storage unit 159 stores information related to reference data. The reference information storage unit 159 includes an initial reference data storage unit (not shown) and a reference distribution information storage unit (not shown). The initial reference data storage unit stores data of a plurality of initial reference images used as an initial reference in structure estimation. The reference distribution information storage unit stores information related to the distribution status of a plurality of reference data sets. In the reference distribution information storage unit, association information indicating which reference set is stored in the data processing unit 113 of which data processing terminal 105 is stored. Specifically, the identifier of the data processing terminal 105 and the identifier of the reference set are stored in association with each other.

  The measurement data arrangement information storage unit 161 stores information regarding the arrangement location of measurement data. The measurement data is stored in the user terminal 101, for example.

  The calculation information storage unit 163 stores the classification data created by the measurement data classification unit 157. Moreover, the arithmetic expression used for the calculation in the structure estimation part 125 may be stored. Further, the calculation information storage unit 163 may store information related to the number of updates. The information regarding the number of updates is, for example, the correspondence between the required number of updates of reference data and an identifier indicating the type of sample.

  Next, a structure estimation procedure using the structure estimation system 100 will be described. The structure estimation is performed based on the single particle analysis method described above with reference to FIG. The data processing (i) and (ii) described above are performed by distributed processing by the plurality of data processing units 113, respectively.

The structure estimation procedure using the structure estimation system 100 is roughly divided into the following four stages.
(1) Structural analysis request-creation of reference data set (2) Comparison between measurement data and reference data set-update of reference (3) repetition (4) structure estimation Hereinafter, for each stage, FIGS. This will be described with reference to FIGS. FIG. 5 is a diagram showing an analysis procedure in the scheduler 103 and the data processing terminal 105. FIG. 6 is a diagram showing in detail a procedure for acquiring a reference data set in the procedure of FIG.

(1) Structural Analysis Request to Reference Data Set Creation First, the user terminal 101 receives a user request and requests structural analysis of the molecule based on a plurality of measurement images. Then, as shown in FIG. 5, the scheduler 103 receives a request for structural analysis of the user terminal 101 (S <b> 101), and requests data processing from the data processing system 110 composed of a plurality of data processing terminals 105. At this time, the scheduler 103 inquires each data processing terminal 105 about the operating status (S103). Based on the report (S201) from each data processing terminal, the scheduler 103 determines a terminal from which a reference data set is acquired (S105), and distributes initial reference data to the corresponding terminal. Also, the scheduler 103 determines a terminal that performs measurement image data processing (S105), sends information related to the distribution state of the measurement image data to the corresponding terminal, and a plurality of reference data included in the data processing system 110. The storage unit 111 distributes data of a plurality of reference images. In the plurality of reference data storage units 111 to which reference image data is distributed, a reference data set is acquired for the reference image distributed to itself.

  In the following, first, a reference data set acquisition procedure will be described with reference to FIGS. 6 and 1. Thereafter, returning to FIG. 5, the procedure of the data processing of the measurement image will be described.

  As shown in FIG. 6, the scheduler 103 confirms the operating status of the data processing terminal 105 (S151), and according to the operating status (S251) reported from the data processing terminal 105, the reference data distribution terminal and the reference A data creation terminal is determined (S153, S157). A data processing terminal 105 to which reference data is distributed has a reference data storage unit 111, and a terminal in charge of creation of reference data has a data processing unit 113.

  Next, the scheduler 103 distributes a plurality of initial reference data (initial reference data) to reference data distribution terminals (S155). At this time, the scheduler 103 communicates with the data processing terminal 105 and distributes the initial reference data to the data processing terminal 105 according to the reception speed of the initial reference data while confirming the operation status reported from the data processing terminal 105. (S155). For example, the scheduler 103 transmits initial reference data to a single data processing terminal 105 in a plurality of times. The scheduler 103 may distribute reference data to some of the plurality of data processing terminals 105 having the reference data storage unit 111. The scheduler 103 may distribute one reference data to the reference data storage units 111 of the plurality of data processing terminals 105.

  For example, in FIG. 1, a data processing terminal 105a, a data processing terminal 105b, and a data processing terminal 105d having a reference data storage unit 111 are selected as terminals for distributing reference data. The scheduler 103 sends the ref. a to c are distributed to the data processing terminal 105b. d to f are distributed to the data processing terminal 105d. Distribute data of g to i.

  In step 155, the data processing terminal 105 having the data processing unit 113 is requested to create a reference data set (S159). For example, in FIG. 1, ref. a to ref. c is requested to create each reference data set, and the ref. d-ref. f is requested to create a reference data set, and the ref. g-ref. Requests creation of i reference data set.

  The data processing terminal 105 that has received the request for creating the reference data set selects the reference data set acquisition method by the reference data set acquisition method selection unit 143 (S255). The reference data set acquisition unit 145 creates the rotationally moving image data and the reference data set by the selected method (S261), and stores them in the reference data set storage unit 139. For example, the reference data set acquisition unit 145 acquires the reference image data stored in the reference image data storage unit 137 (S257), and each of them rotates, for example, by rotating from 1 degree to 360 degrees at a pitch of 1 degree. Image group data is created (S259). Then, a reference data set including the data of the rotationally moving image group and the image rotated by zero degrees, that is, the data of the reference image is created (S261).

  Note that in step 259, data of a rotationally moving image group of up to 90 degrees can be created without creating data of a rotationally moving image group of up to 360 degrees, for example. At this time, in the similarity parameter calculation procedure to be described later, the data of the rotationally moving image group up to 360 degrees can be created by the sign inversion of the data of the rotationally moving image group up to 90 degrees. Therefore, the creation amount of the reference data set in step 159 and the data amount stored in the reference data set storage unit 139 are reduced to speed up the creation process of the reference data set, and the calculation process for the similarity parameter is also efficient. Can be done well.

  In the example shown in FIG. 1 and FIG. 3, for example, the data processing unit 113 of the data processing terminal 105 a receives the ref. Data stored in its own reference image data storage unit 137. a to ref. Referring to c, a rotationally moving image group is created for each reference image. As a reference data set, ref. seta and ref. get setb. The obtained reference data set is stored in its own reference data set storage unit 139. For the data processing terminal 105b and the data processing terminal 105d, a reference data set is created using the same method as the data processing terminal 105a.

  Note that the data processing terminal 105 does not necessarily have to create a reference data set for reference stored in its own reference data storage unit 111, and the scheduler 103 creates a reference image for which a reference data set is to be created and the reference image. The data arrangement status may be sent to the reference data set creation terminal. In this way, the reference data set can be created more quickly according to the operating status of the data processing terminal 105.

(2) Comparison of Measurement Data and Reference Data Set to Reference Update Returning to FIG. 5, the scheduler 103 includes a plurality of measurement images, a plurality of reference images, and a plurality of data processing units 113 included in the data processing system 110. Assign a process to evaluate the similarity of. Then, the scheduler 103 presents the measurement data arrangement information stored in the measurement data arrangement information storage unit 161 and the identifier of the reference image to be data processed to the data processing terminal 105 that performs data processing of the measurement image. Then, data processing is requested (S107). For example, in the example of FIG. 1, each of the data processing terminal 105a to the data processing terminal 105e has ref. ab, ref. cd, ref. ef, ref. g to h and ref. Request data processing for i. The data processing terminal 105 that has received the request acquires the presented measurement data arrangement information (S203).

  Each data processing terminal acquires a reference data set of a reference image in charge as an initial reference (S205). The acquisition method is as described above with reference to FIG. 6, and the data processing terminal 105 may create a reference data set by itself or store data of the reference data set of the reference image in charge. Data may be acquired from the reference data storage unit 111 of the terminal. Also, after communicating with other data processing terminals 105, the response time is used to compare the time required to receive the reference data set with the time required to create the reference data set, and determine whether to create the reference data set. May be. When a reference data set is acquired from another data processing terminal 105 and the data of the target reference set is stored in a plurality of data processing terminals 105, from which data processing terminal 105 the data is acquired Also determine whether is the fastest. Further, when a reference data set of a target reference is stored in its own reference data storage unit 111 as in the data processing terminal 105a of FIG. 1, it can be acquired and used.

  In addition, the data processing terminal 105 obtains measurement data based on the measurement data arrangement information at the similarity parameter assignment unit 147 (S207). For example, when the measurement data is stored in the first storage unit 107 of the user terminal 101, the stored measurement data is acquired. In addition, the data processing terminal 105 acquires a translation image of the measurement data (S209). At this time, the data processing terminal 105 creates translation image data by referring to the measurement data stored in the measurement image data storage unit 133 and the arithmetic expression stored in the translation image information storage unit 135. Also good. Moreover, you may acquire the data of the translation image stored in own measurement image data storage part 133 or another terminal.

  Then, the data processing terminal 105 includes the measurement image data or the parallel movement image data obtained by translating the measurement image, and the assigned reference image data and the rotational movement image group obtained by rotating the reference image. A reference data set composed of data is compared, and processing for evaluating the similarity between the measurement image and the assigned reference image is performed. The data processing terminal 105 compares the reference data set of the reference in charge with the measurement data and its translation image, calculates a parameter indicating the similarity between the measurement data and the reference, and determines the identifier of the measurement data, the reference The identifier and the similarity parameter are associated with each other, stored in the similarity information storage unit 151, and transmitted to the scheduler 103 (S211).

  The scheduler 103 associates each of the plurality of measurement images with one of the plurality of reference images based on the result of the process for evaluating the similarity. At this time, the scheduler 103 acquires the measurement data with the similarity parameter sent from each data processing terminal 105 (S109), extracts the reference with the highest similarity for each measurement data, and extracts the measurement data A classification parameter is assigned to the identifier, stored in the calculation information storage unit 163, and sent to the data processing terminal 105 (S111). The classification parameter is, for example, an identifier of a reference image having the highest similarity.

  The data processing terminal 105 acquires reference image data to which the classification parameter is assigned (S213). Then, the averaging processing unit 149 refers to the data of a plurality of measurement images to which the identifier of the reference image having the highest similarity is assigned, and averages the measurement images associated with the same reference (S215). Then, the reference image data storage unit 137 is updated with the averaged image data obtained by the averaging as the new reference image data (S217).

(3) Repeat In the above procedure, the steps from the acquisition of the reference data set in step 205 to the reference update in step 217 are repeated a predetermined number of times (YES in S219).

(4) Structure estimation After updating the reference repeatedly a predetermined number of times (NO in S219), the data processing terminal 105 presents the updated reference image data to the scheduler 103 (S221). Then, the scheduler 103 acquires data of a plurality of averaged images obtained by averaging a plurality of measurement images associated with any of the plurality of reference images, and based on the data of the averaged images, the numerator Estimate the three-dimensional structure. At this time, the structure estimation unit 125 creates estimation data of the three-dimensional structure of the membrane protein to be analyzed based on the arithmetic expression stored in the calculation information storage unit 163 and the acquired reference image data (S113), and the user terminal 101 (S115).

  According to this embodiment, when estimating the three-dimensional structure of a biopolymer by a single particle analysis method, reference set data (reference data set) composed of a plurality of reference images and a rotationally moving image group is preliminarily stored. It is stored in the reference data storage unit 111. Conventionally, without creating such a reference data set in advance, a parallel movement image of a measurement image and a rotational movement image corresponding to each of the measurement images are created and aligned with the reference image. I was evaluating. For this reason, the load of data processing is high, the load is concentrated on the terminal that performs data processing, and a long time is required for analysis. On the other hand, in the present embodiment, since the data processing system 110 holds the reference set data in advance, it is easy to calculate the similarity parameter, and it is necessary to evaluate the similarity while rotating the measurement image. There is no. For this reason, data processing can be performed efficiently and quickly. In addition, even when the user terminal 101 does not have a data processing unit, it is possible to request structure estimation via the network 115 and quickly obtain an analysis result.

  Furthermore, in the present embodiment, the reference data set is created by distributed processing by the plurality of data processing units 113, and also when the reference data set and the measurement image data are compared, the plurality of data processing units 113 Distributed processing is performed. For this reason, it can suppress that the load of the predetermined | prescribed data processing part 113 in the data processing system 110 concentrates, and the creation process of a reference set can be performed efficiently and rapidly.

  In addition, the scheduler 103 includes a data processing unit selection unit 155. Therefore, the data processing unit selection unit 155 selects a data processing terminal 105 having a short distance on the network, a data processing terminal 105 having a data processing unit 113 with a light load, a data processing terminal 105 having a wide communication bandwidth, and the like. Requesting distributed processing and distributing data. It is also possible to request the data processing unit 113 of the data processing terminal 105 to compare with the reference data set stored in the reference data storage unit 111 of the same data processing terminal 105. For this reason, the processing speed can be further improved.

As described above, in the structure estimation system 100, a plurality of processes (i) creation of the reference data set and (ii) evaluation of similarity between the measurement image and the reference data set, which have a large load in data processing, are performed. The distributed processing by the data processing unit 113 is applied. Reference image data is distributed to a plurality of data processing terminals 105. In addition, a plurality of data processing terminals 105 share the reference image set for the reference image. Therefore, it is possible to efficiently and quickly perform data processing using a large amount of measurement data and a large amount of reference data set without placing a load on the user terminal 101 or the specific data processing terminal 105.

  Further, in the present embodiment, the similarity between the reference data set and the measurement image is evaluated, and a similarity parameter is assigned to the identifier of the measurement image, so that a plurality of measurement data is referred to as the reference having the highest similarity. It is possible to classify in association with an image identifier efficiently. For this reason, analysis using a plurality of measurement images having different measurement directions can be efficiently performed.

  In this embodiment, the averaged image obtained by averaging the measurement data classified for each reference is updated as a new reference image, and the update process is repeated. For this reason, it is difficult to crystallize using conventional methods, such as membrane proteins, so even for samples for which three-dimensional structure analysis data has not accumulated, update work using a predetermined image as an initial reference. By repeating the above, it is possible to bring the reference image closer to the actual image. In addition, the structure estimation system 100 can be effectively used for analysis of biopolymers other than membrane proteins, for example. For example, the structure estimation system 100 may be used for analyzing the three-dimensional structure of polysaccharides and glycoproteins.

(Second embodiment)
In the first embodiment, the configuration in which the reference data set acquisition method selection unit 143 is provided in the data processing terminal 105 is illustrated. However, in the structure estimation system 100 illustrated in FIG. 2, the reference data set acquisition method selection unit 143 May be provided in the scheduler 103. In the present embodiment, such a configuration will be described.

  FIG. 4 is a functional block diagram showing details of the scheduler 103 and the data processing terminal 105 of the structure estimation system of this embodiment. In FIG. 4, the scheduler 103 includes a reference data set acquisition method selection unit 143, and a plurality of data processing units 113 receives a reference data set acquisition method selection reception in which the reference data set acquisition unit 145 selects a reference data set acquisition method. Part 165.

  In this embodiment, the reference data set acquisition method selection unit 143 of the scheduler 103 grasps the operating status of the data processing terminal 105 and selects a reference data set acquisition method. Then, information on the acquisition method is presented to the data processing terminal 105 that performs data processing. The data processing terminal 105 receives information on the acquisition method at the reference data set acquisition method selection reception unit 165. The reference data set acquisition unit 145 acquires or creates a reference data set based on the acquisition method received by the reference data set acquisition method selection reception unit 165.

  In this embodiment, the scheduler 103 grasps the operating status of the entire data processing terminal group, and instructs the data processing terminal 105 that performs data processing to acquire a reference data set. Can be performed more efficiently.

  As mentioned above, although embodiment of this invention was described with reference to drawings, these are the illustrations of this invention, Various structures other than the above are also employable.

  For example, in the above embodiment, the case where the averaging processing unit 149 is provided in the data processing terminal 105 has been described as an example, but the scheduler 103 may include the averaging processing unit 149. Moreover, the averaging processing unit 149 may be provided in both the scheduler 103 and the data processing terminal 105. At this time, the scheduler 103 confirms the operation status of the data processing terminal 105 and determines whether the data processing terminal 105 performs the averaging process or whether the averaging process unit 149 performs the averaging process. There may be provided a processing method determination unit.

  Further, in the above embodiment, the scheduler 103 may be configured by a plurality of terminal devices, and each functional block may be distributed and arranged in a plurality of terminal devices. In addition, the user terminal 101 may be configured from a plurality of terminals.

  In the above embodiment, the substance to be analyzed may be composed of a single molecule, or a plurality of molecules may form a supramolecular structure.

It is a figure explaining the structure of the structure estimation system which concerns on this embodiment. It is a functional block which shows the structure of the structure estimation system which concerns on this embodiment. It is a functional block diagram which shows the detail of the scheduler and data processing terminal of the structure estimation system of FIG. It is a functional block diagram which shows the detail of the scheduler and data processing terminal of the structure estimation system of FIG. It is a figure explaining the structure estimation procedure which concerns on this embodiment. It is a figure explaining the structure estimation procedure which concerns on this embodiment. It is a figure explaining a single particle analysis method.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Structure estimation system 101 User terminal 103 Scheduler 105 Data processing terminal group 105a Data processing terminal 105b Data processing terminal 105c Data processing terminal 105d Data processing terminal 105e Data processing terminal 107 1st memory | storage part 109 2nd memory | storage part 111 Reference data memory | storage part 113 Data processing unit 115 Network 117 First transmission / reception unit 119 Second transmission / reception unit 121 Third transmission / reception unit 123 Operation unit 125 Structure estimation unit 127 Measurement data classification unit 129 Third storage unit 131 Measurement data storage unit 133 Measurement image data storage unit 135 Parallel Moving image information storage unit 137 Reference image data storage unit 139 Reference data set storage unit 141 Parallel moving image acquisition unit 143 Reference data set acquisition method selection unit 145 Reference data set acquisition unit 147 Similarity parameter assignment unit 149 Averaging processing unit 151 Similarity information storage unit 153 Averaged information storage unit 155 Reference data update unit 157 Measurement data classification unit 159 Reference information storage unit 161 Measurement data arrangement information storage unit 163 Calculation information storage unit 165 Reference data set acquisition method selection reception unit

Claims (11)

A system for estimating the three-dimensional structure of a molecule from data of a plurality of images,
A user terminal that accepts a user request and requests a structural analysis based on a plurality of measurement images of the molecule;
A data processing system comprising a plurality of data processing terminals;
A scheduler that accepts a request from the user terminal and requests data processing from the data processing system;
Are connected via a network,
The data processing system is
A plurality of reference data storage units to which data of a plurality of reference images are distributed;
A plurality of data processing units;
Including
Each of the data processing units translates the measurement image data or the translation image data obtained by translating the measurement image for each of the plurality of measurement images in response to the data processing request from the scheduler. And a reference data set composed of data of the reference image assigned to itself and data of a group of rotationally moving images obtained by rotating and moving the reference image, and assigned to the measurement image. There line processing for evaluating the similarity between the reference image,
The scheduler
A measurement data classifying unit that associates a plurality of the measurement images with any one of the plurality of reference images based on the result of the processing for evaluating the similarity;
The averaged image data obtained by averaging the plurality of measurement images associated with the same reference image in the measurement data classifying unit is acquired for the plurality of reference images, and the plurality of averaged images acquired A structure estimation unit for estimating the three-dimensional structure of the molecule based on
Structure estimation system including The structure estimation system according to claim 1,
A plurality of the averaged image data is redistributed as updated reference images to the plurality of reference data storage units, and the updated reference images are allocated to the plurality of data processing units,
The data processing unit performs a process of evaluating the similarity between the measurement image and the assigned update reference image,
The measurement data classifying unit associates the plurality of measurement images with any of the plurality of updated reference images based on the result of the process of evaluating the similarity,
The structure estimation unit acquires data of an averaged image obtained by averaging the plurality of measurement images associated with the same updated reference image in the measurement data classification unit for the plurality of updated reference images, A structure estimation system that estimates the three-dimensional structure of the molecule based on the plurality of acquired averaged images. The structure estimation system according to claim 1 or 2,
The scheduler
A data processing unit selection unit that selects a plurality of data processing units to which a plurality of reference images are assigned for the processing for evaluating similarity based on operating states of the plurality of data processing units;
A reference distribution information storage unit for storing information on a distribution status of the plurality of reference data sets;
Structure estimation system including   The structure estimation system according to claim 3, wherein the data processing unit selection unit further selects a plurality of the data processing units that acquire the reference data set. The structure estimation system according to any one of claims 1 to 4,
A plurality of the data processing units,
A reference data set acquisition unit for acquiring the reference data set;
A reference data set acquisition method selection unit that selects an acquisition method of the reference data set in the reference data set acquisition unit;
Structure estimation system including The structure estimation system according to any one of claims 1 to 4,
A plurality of the data processing units include a reference data set acquisition unit that acquires the reference data set,
The structure estimation system in which the scheduler includes a reference data set acquisition method selection unit that selects an acquisition method of the reference data set in the reference data set acquisition unit.   7. The structure estimation system according to claim 1, wherein the reference data set includes data of the reference image and data of a rotationally moving image group obtained by rotating the reference image from 0 degrees to 90 degrees. Structure estimation system.   8. The structure estimation system according to claim 1, wherein the molecule is a biopolymer.   The structure estimation system according to claim 8, wherein the biopolymer is a membrane protein. A method for estimating the three-dimensional structure of a molecule from data of a plurality of images,
A user terminal accepting a user request and requesting a structural analysis of the molecule based on a plurality of measurement images;
A scheduler accepting a request from the user terminal and requesting data processing from a data processing system including a plurality of data processing terminals;
A step wherein the scheduler, the multiple's Reference data storage unit included in the data processing system, for distributing the data of a plurality of reference images,
A step wherein the scheduler is to allocate the data processing plurality of data processing units included in the system, the process of evaluating the similarity between a plurality of measurement images and the plurality of reference images,
Each of the plurality of data processing units is assigned to the data of the measurement image or the data of the translation image obtained by translating the measurement image in response to the data processing request from the scheduler . by comparing the reference data set composed of a data group of rotational movement image composed by rotating moving data and the reference image of the reference image, similar to the reference image allocated to the measurement image Performing the process of evaluating the degree;
The scheduler associates each of the plurality of measurement images with one of the plurality of reference images based on the result of the process of evaluating the similarity,
The scheduler acquires a plurality of averaged image data obtained by averaging the plurality of measurement images associated with any of the plurality of reference images, and based on the averaged image data, Estimating the conformation of the molecule;
A structure estimation method including: The computer,
A system for estimating the three-dimensional structure of a molecule from data of a plurality of images,
A user terminal that accepts a user request and requests a structural analysis based on a plurality of measurement images of the molecule;
A data processing system comprising a plurality of data processing terminals;
A scheduler that accepts a request from the user terminal and requests data processing from the data processing system;
Are connected via a network,
The data processing system is
A plurality of reference data storage units to which data of a plurality of reference images are distributed;
A plurality of data processing units;
Including
Each of the data processing units translates the measurement image data or the translation image data obtained by translating the measurement image for each of the plurality of measurement images in response to the data processing request from the scheduler. And a reference data set composed of data of the reference image assigned to itself and data of a group of rotationally moving images obtained by rotating and moving the reference image, and assigned to the measurement image. There line processing for evaluating the similarity between the reference image,
The scheduler
A measurement data classifying unit that associates a plurality of the measurement images with any one of the plurality of reference images based on the result of the processing for evaluating the similarity;
The averaged image data obtained by averaging the plurality of measurement images associated with the same reference image in the measurement data classifying unit is acquired for the plurality of reference images, and the plurality of averaged images acquired A structure estimation unit for estimating the three-dimensional structure of the molecule based on
A program that functions as a structure estimation system.

Images