KR102619707B1 - Method and system for processing bio data based on distributed parallel processing - Google Patents

Abstract

A system for distributed and parallel processing of bio data is disclosed. This system receives a 3D data set of a certain capacity including EM data corresponding to each of the first data type and the second data type, and based on the data type information included in the 3D data set, the first data type It includes a master node that classifies the EM data corresponding to the EM data and the EM data corresponding to the second data type, and one or more operation nodes clustered with the master node. Accordingly, processing of large amounts of image data can be performed.

Description

Method and system for distributed parallel processing of bio data {METHOD AND SYSTEM FOR PROCESSING BIO DATA BASED ON DISTRIBUTED PARALLEL PROCESSING}

The present invention relates to a method of distributed parallel processing of bio data and a system to which the same is applied. More specifically, to a method of distributed parallel processing of a three-dimensional data set observed with an electron microscope and a system to which the same is applied.

In the field of biological research, with the development of next-generation sequencing technology, image-based activity data, genome data, transcriptome data, and proteome data are being generated, and these data are characterized as large data due to the characteristics of high resolution and other characteristics. .

In addition, technology for imaging biological samples continues to develop. Typically, an electron microscope is a microscope that uses an electron beam as a light source, and can observe various samples such as photonic crystals, protein molecules, cells, and cellular tissues. In the case of proteins, the samples are observed at ultra-low temperature to obtain atomic-level resolution. Structural analysis can be performed.

Since the 3D image data observed and generated by an electron microscope is gigabyte or terabyte-level data, it was difficult to process and process such large data in the prior art.

Meanwhile, the above information is presented only as background information to aid understanding of the present invention. No decision has been made, and no claim is made, as to whether any of the above is applicable as prior art with respect to the present invention.

Publication of Patent No. 10-2016-0099762 (Publication date: 2016.08.23)

The present invention was devised to solve the above-mentioned problems, and one object of the present invention is to provide a distributed parallel processing method for image processing a 3D image data set of an electron microscope.

Another object of the present invention is to provide a method of processing a 3D image data set into a plurality of 3D chunks.

The technical problems to be achieved in the present invention are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the description below. You will be able to.

A method of distributed parallel processing of bio data according to an embodiment of the present invention for realizing the above-mentioned problem is 3 of a predetermined capacity including EM (Electron Microscope) data corresponding to each of the first data type and the second data type. Receiving a dimensional data set; Classifying EM data corresponding to the first data type and EM data corresponding to the second data type based on data type information included in the 3D data set; Based on key and value information corresponding to each of the first data type and the second data type, a first message queue corresponding to the first data type and a second message corresponding to the second data type sequentially providing computational tasks to a queue; creating a semaphore for allocating shared resources to the first message queue and the second message queue, and determining a computation node to perform the computation tasks provided sequentially; And it may include the step of performing the computational task by the determined computational node.

A system for distributed parallel processing of bio data according to an embodiment of the present invention receives a three-dimensional data set of a certain capacity including EM (Electron Microscope) data corresponding to each of the first data type and the second data type, a master node that classifies EM data corresponding to the first data type and EM data corresponding to the second data type based on data type information included in the three-dimensional data set; And it may include one or more computational nodes clustered with the master node.

The master node generates a first message queue corresponding to the first data type and a second message queue corresponding to the second data type, based on key and value information corresponding to each of the first data type and the second data type. Computation tasks may be provided sequentially, a semaphore for allocating shared resources to the first message queue and the second message queue may be created, and a computation node to perform the sequentially provided computation tasks may be determined. The determined computational node may be configured to perform computational tasks.

The technical solutions to be achieved in the present invention are not limited to the technical solutions mentioned above, and other technical solutions not mentioned above will be clear to those skilled in the art from the description below. It will be understandable.

According to various embodiments of the present invention, a large amount of 3D image data observed by an electron microscope can be quickly processed through distributed parallel processing to generate a plurality of 3D chunk data.

The technical effects to be achieved in the present invention are not limited to the technical effects mentioned above, and other technical effects not mentioned will be clearly understood by those skilled in the art from the description below. You will be able to.

1 is a diagram schematically illustrating a system for distributed and parallel processing of bio data according to an embodiment of the present invention;
FIG. 2 is a diagram illustrating a clustered master node and a plurality of operation nodes of a distributed parallel processing system according to an embodiment of the present invention;
3 is a diagram illustrating a master node process and a calculation node process of a distributed parallel processing system according to an embodiment of the present invention, and
Figure 4 is a sequence diagram showing a method of distributed parallel processing of bio data according to an embodiment of the present invention.

The advantages and features of the present invention and methods for achieving them will become clear by referring to the embodiments described in detail below along with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and will be implemented in various different forms. The present embodiments only serve to ensure that the disclosure of the present invention is complete and that common knowledge in the technical field to which the present invention pertains is not limited. It is provided to fully inform those who have the scope of the invention, and the present invention is only defined by the scope of the claims.

The shapes, sizes, proportions, angles, numbers, etc. disclosed in the drawings for explaining embodiments of the present invention are illustrative, and the present invention is not limited to the matters shown. Like reference numerals refer to like elements throughout the specification.

Additionally, in describing the present invention, if it is determined that a detailed description of related known technologies may unnecessarily obscure the gist of the present invention, the detailed description will be omitted.

When 'includes', 'has', 'consists of', etc. mentioned in this specification are used, other parts may be added unless 'only' is used. When a component is expressed in the singular, the plural is included unless specifically stated otherwise.

When interpreting a component, it is interpreted to include the margin of error even if there is no separate explicit description. In the case of a description of a positional relationship, for example, if the positional relationship of two parts is described as 'on top', 'on the top', 'on the bottom', 'next to', etc., 'immediately' Alternatively, there may be one or more other parts placed between the two parts, unless 'directly' is used.

In the case of a description of a temporal relationship, for example, if a temporal relationship is described as 'after', 'successfully after', 'after', 'before', etc., 'immediately' or 'directly' Unless used, non-consecutive cases may also be included.

Although first, second, etc. are used to describe various components, these components are not limited by these terms. These terms are merely used to distinguish one component from another. Accordingly, the first component mentioned below may also be the second component within the technical spirit of the present invention.

“X-axis direction,” “Y-axis direction,” and “Z-axis direction” should not be interpreted as only geometrical relationships in which the relationship between each other is vertical, and should not be interpreted as a wider range within which the configuration of the present invention can function functionally. It can mean having direction.

The term “at least one” should be understood to include all possible combinations from one or more related items. For example, “at least one of the first, second, and third items” means each of the first, second, or third items, as well as two of the first, second, and third items. It can mean a combination of all items that can be presented from more than one.

Each feature of the various embodiments of the present invention can be combined or combined with each other, partially or entirely, and various technical interconnections and operations are possible, and each embodiment can be implemented independently of each other or together in a related relationship. It may be possible.

FIG. 1 is a diagram schematically illustrating a system for distributed parallel processing of bio data (hereinafter referred to as “distributed parallel processing system”) according to an embodiment of the present invention.

Referring to FIG. 1, the electron microscope 10 can photograph a sample. Here, the sample 17 may be extracted from various parts of the living body. As an example, sample 17 may be extracted from the brain of an organism.

The electron microscope 10 outputs an electron beam 15 using an electron gun 11, and the output electron beam 15 passes through a magnetic lens 13 to obtain a sample. The image for (17) can be output on the screen (Screen, 19).

The image processing module 20 can process captured images in a separate device or in the cloud. Specifically, the image processing module 20 performs a stitch process and an alignment process on numerous two-dimensional images captured from the electron microscope 10, generating the two-dimensional images into three-dimensional images. can do.

The image processing module 20 may provide the generated three-dimensional (3D) image to the distributed parallel processing system 100. The generated 3D image can be implemented as a single file, and can also be implemented with a capacity of hundreds of gigabytes to tens of terabytes.

The distributed parallel processing system 100 may be implemented as a cloud, but in an optional embodiment, it may also be implemented as a server or device.

The distributed parallel processing system 100 may receive a three-dimensional (3D) image from the image processing module 20 and process the received image. In an optional embodiment, the distributed parallel processing system 100 may directly receive a 3D image through a specific input means. The specific configuration and operation of the distributed parallel processing system 100 will be described in detail below.

Figure 2 is a diagram for explaining the clustered master node 110 and a plurality of operation nodes 1501 to 150N of the distributed parallel processing system 100 according to an embodiment of the present invention.

Referring to FIG. 2, the master node 110 and a plurality of operation nodes 1501 to 150N of the distributed parallel processing system 100 may form a group on a cluster (CLU) basis.

Figure 3 is a diagram for explaining the master node process (S31) and the operation node process (S33) of the distributed parallel processing system 100 according to an embodiment of the present invention.

The master node process (S31) refers to a process performed by the master node 110 or a module directly or indirectly connected to the master node 110, and is described as being performed by the master node 110. The node process S33 refers to a process performed by the computation node 150 or a module directly or indirectly connected to the computation node 150, and will be described as being performed by the computation node 150.

The master node process (S31) may include a master node 110, a channel & segmentation classification module 120, a plurality of message queues (130A, 130B), an error report module, a thread safe queue, etc.

First, the master node process (S31) can receive an EM (Electron Microscope) data set. Specifically, the master node 110 may receive an EM data set and provide the data set to the channel & segmentation classification module 120.

Here, the EM data set can be implemented in the form of a single file containing 3D data. The EM data set may include multiple EM data (sets) that can be divided into multiple data types.

For example, an EM data set may be a 3D image data set collected from biological samples located in the brain of humans, animals, etc., and a 3D image data set may be a single file containing 3D image data (sets) expressed in different data types. It can be stored together inside. The extension of a single file may be <.H5>. Here, the size of a single file can range from gigabytes to terabytes.

Here, the 3D-based EM data set may include EM data (sets) corresponding to each of the first data type and the second data type. The first data type may be a Channel type, and the second data type may be a Segmentation type. A channel is an image that expresses a 3D image data set in gray scale for each pixel, and segmentation is an image that expresses a 3D image data set in color scale for each pixel.

In addition, the EM data corresponding to the first data type (channel type) is EM data expressed based on gray scale for the EM data set, and is EM data expressed based on gray scale for each pixel of the EM data set expressed in three dimensions. am. Pixels with bright luminance can be expressed as white, and pixels with dark luminance can be expressed as black.

EM data corresponding to the second data type (segmentation) is EM data expressed based on a color scale for the EM data set, and may include a color pixel value for each pixel of the EM data set expressed in three dimensions. For example, when a color pixel value is expressed in UNIT16 size, a color value in the range of 0 to 65535 can be set per pixel.

In other words, the EM data set according to an embodiment of the present invention consists of 3D EM data expressed in multiple types as one file, and prior art does not consist of 3D EM data expressed in multiple types as one file. It can be said that it has overcome the limitations of

The master node process (S31) uses the channel & segmentation classification model 120 to generate EM data corresponding to the first data type (channel type) and the first data type based on the data type information included in the 3D-based EM data set. 2 EM data corresponding to the data type (segmentation type) can be classified.

At this time, the master node process (S31) can sequentially deliver jobs for calculation to a Thread Safe Queue message, and the master node 110 and /Or it can be monitored through a task manager, etc.

The master node process (S31) processes the first message queue 130A and the first message queue 130A corresponding to the first data type based on key and value information corresponding to each of the first data type and the second data type. 2 Computation tasks may be sequentially provided to the second message queue 130B corresponding to the data type.

Here, the key and value information are memory-based information for storing a large 3D EM data set. The key information may include file extension (H5) information or information related to the extension information, and the value information is It may include information about the first data type (channel type) or information about the second data type (segmentation type).

Additionally, the computation tasks loaded in the message queues 130A and 130B may be computation tasks for generating a 3D EM data set into 3D EM data chunks of a predetermined size.

The master node process S31 creates a semaphore to allocate shared resources to the first message queue 130A and the second message queue 130B, thereby preventing duplicate processing or deadlock.

The master node process (S31) can determine the calculation nodes (1501 to 150N) that will sequentially perform the provided calculation tasks.

After the master node process (S31) is performed, the determined calculation nodes (150, 1501 to 150N) can perform calculation tasks. Specifically, the operation nodes 150 and 1501 to 150N may obtain permission to use shared resources of the first message queue 130A or the second message queue 130B through a semaphore. The operation nodes 150 may perform both operations related to the first data type and operations related to the second data type.

In an embodiment, more data types than the first and second data types may be applied, and in this case, the number of message queues may increase depending on the data type.

Computation nodes 1501 to 150N may generate a 3D data set corresponding to the computation task as 3D chunk data. A 3D chunk may be 3D image data obtained by dividing a large 3D data set into predetermined sizes. For example, based on three-dimensional

The distributed parallel processing system 100 may further include a display, and output the three-dimensional chunks generated by the operation nodes 1501 to 150N to the display.

Additionally, the master node process 31S may receive operation error log information when an operation error occurs in the operation node.

Based on the operation error log information, the master node process 31S can ignore relatively light operation error logs and stop the process when a predetermined level of error log is found.

The master node process 31S includes an artificial intelligence-based error processing model and can automatically stop distributed parallel processing when an error log exceeding a predetermined standard is found.

Figure 4 is a sequence diagram showing a method of distributed parallel processing of bio data according to an embodiment of the present invention.

First, the processing method of the bio data distributed parallel processing system 100 includes the step of receiving a three-dimensional data set of a predetermined capacity including EM (Electron Microscope) data corresponding to each of the first data type and the second data type (S510) ), may include classifying the EM data corresponding to the first data type and the EM data corresponding to the second data type based on data type information included in the 3D data set (S520).

Next, the processing method is based on key-value information corresponding to each of the first data type and the second data type, to the first message queue and the second data type corresponding to the first data type. Step (S530) of sequentially providing calculation tasks to the corresponding second message queue, creating a semaphore for allocating shared resources to the first message queue and the second message queue, and providing the calculation tasks sequentially provided. It may include a step (S540) of determining the operation node to be performed.

Afterwards, the processing method may include a step (S550) in which the determined operation nodes generate a 3D data set as 3D-based chunk data.

Meanwhile, according to various embodiments of the present invention, by distributing and parallel processing EM data of different data types, distributed processing of 600 gigabytes of EM data takes several hours, which is a limitation of several days according to the prior art. can be improved.

Additionally, according to various embodiments of the present invention, it may be helpful in identifying neural circuits related to decision-making in the frontal lobe. In addition, it can be helpful in the field of brain neural network technology through storage, extraction, and analysis technology and visualization technology of brain neural network image data, structural data, and molecular data.

Although this specification contains details of numerous specific implementations, these should not be construed as limitations on the scope of any invention or what may be claimed, but rather as descriptions of features that may be unique to particular embodiments of particular inventions. It must be understood. Likewise, certain features described herein in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination. Furthermore, although features may be described as operating in a particular combination and initially claimed as such, one or more features from a claimed combination may in some cases be excluded from that combination, and the claimed combination may be a sub-combination. It can be changed to a variant of a sub-combination.

Additionally, although operations are depicted in the drawings in a specific order herein, this should not be understood to mean that such operations must be performed in the specific order or sequential order shown or that all illustrated operations must be performed to obtain desirable results. Can not be done.

As such, this specification is not intended to limit the invention to the specific terms presented. Accordingly, although the present invention has been described in detail with reference to the above-described examples, those skilled in the art may make modifications, changes, and variations to the examples without departing from the scope of the present invention. The scope of the present invention is indicated by the claims described below rather than the detailed description above, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present invention. do.

Claims (12)

As a method for distributed parallel processing of bio data,
Receiving a three-dimensional data set of a certain capacity including EM (Electron Microscope) data corresponding to each of a gray scale-based first data type and a color scale-based second data type;
Based on the data type information included in the three-dimensional data set, classifying the EM data included in the three-dimensional data set into EM data corresponding to the first data type and EM data corresponding to the second data type. step;
Based on key and value information corresponding to each of the first data type and the second data type, a first message queue corresponding to the first data type and a second message corresponding to the second data type sequentially providing computational tasks to a queue;
creating a semaphore for allocating shared resources to the first message queue and the second message queue, and determining a computation node to perform the computation tasks provided sequentially; and
Distributed parallel processing method comprising the step of the determined computational node performing a computational task. According to paragraph 1,
The 3D data set is comprised of a single file, and the 3D data set is comprised of a capacity of gigabytes to terabytes. delete According to paragraph 1,
The step of performing the calculation task is,
A distributed parallel processing method comprising generating a 3D data set corresponding to the computational task as 3D chunk data. According to paragraph 4,
Distributed parallel processing method further comprising displaying the generated 3D chunk data. According to paragraph 1,
Distributed parallel processing method further comprising receiving operation error log information when an operation error occurs in the operation node. As a system for distributed parallel processing of bio data,
Receives a 3D data set of a certain capacity including EM (Electron Microscope) data corresponding to each of a gray scale-based first data type and a color scale-based second data type, and data included in the 3D data set A master node that classifies EM data included in the three-dimensional data set into EM data corresponding to the first data type and EM data corresponding to the second data type, based on type information; and
It includes one or more computational nodes clustered with the master node,
The master node is,
Based on the key and value information corresponding to each of the first data type and the second data type, an operation task (Task) is sent to the first message queue corresponding to the first data type and the second message queue corresponding to the second data type. ) are provided sequentially,
Creates a semaphore for allocating shared resources to the first message queue and the second message queue, and determines a computational node to perform the sequentially provided computational tasks,
A distributed parallel processing system in which the determined computational nodes are configured to perform computational tasks. In clause 7,
The 3D data set is comprised of a single file, and the 3D data set is comprised of a capacity of gigabytes to terabytes. delete In clause 7,
The determined operation node is,
A distributed parallel processing system configured to generate a 3D data set corresponding to the computational task as 3D chunk data. According to clause 10,
A distributed parallel processing system further comprising a display for displaying the generated three-dimensional chunk data. In clause 7,
The master node is,
A distributed parallel processing system configured to receive computational error log information when an computational error occurs in a computational node.

Images