Structure determination system, structure determination method, and program

A structure determination system which classifies a plurality of measured images relating to a membrane protein into a plurality of images and determines a structure of said membrane protein based on the classification result, said structure determination system comprising:

• a reference image data storage unit which stores data representing a plurality of reference images;
• a measured image classifying unit which generates data representing a plurality of groups of derived images being derived from said measured images by changing a relative position of each of said measured images with respect to said reference images, evaluates similarity between said derived images and said reference images for each of said plurality of groups of derived images, and extracts, from each of said groups of derived images, a plurality of derived images highly similar to any one of said plurality of reference images;
• an averaged image data generating unit which classifies said plurality of derived images extracted by said measured image classifying unit into a plurality of groups, averages the derived images classified into a common group to generate data representing a plurality of averaged images, and determines whether updating of data representing said reference images is to be performed, said averaged image data generating unit performing the updating of data representing said plurality of reference images in said reference image data storage unit using data representing said plurality of averaged images and sending a request causing said measured image classifying unit to perform processing for extracting said derived images when determining that the updating is necessary, and said averaged image data generating unit setting the generated data representing a plurality of averaged images as data to be used for structure determination when determining that the updating is not necessary;
• and
• a structure determining unit which acquires data representing said plurality of averaged images generated by said averaged image data generating unit and determines a structure of said subject of analysis based on the acquired data,
• wherein:
• the number of pixels of each of said derived images and said reference images is n×m (where n and m are natural numbers);
• on the basis of indexes which represent respective distances between said derived images and said reference images in an n×m-dimensional image space, said measured image classifying unit is configured to search for a derived image which corresponds to a maximum peak of the indexes, as one of said plurality of derived images which is most similar to any one of said plurality of reference images, and for one or more derived images each corresponding to a difference Δ of a peak of the indexes representing said distances, said difference Δ being within a certain range;
• said averaged image data generating unit classifies said plurality of derived images based on positional proximity among said plurality of derived images in an n×m-dimensional image space or an L-dimensional pixel space, and further updates data representing said plurality of reference images in said reference image data storage unit using data representing said plurality of averaged images and sends a request causing said measured image classifying unit to perform processing for extracting the derived images;
• said measured image classifying unit receives said request, evaluates similarity between said derived images and the updated reference images, and extracts, from each of said groups of derived images, a plurality of derived images highly similar to any one of the updated reference images;
• said averaged image data generating unit classifies said plurality of derived images newly extracted by said measured image classifying unit into a plurality of groups on the basis of positional proximity among said plurality of derived images in said L-dimensional pixel space, averages the newly extracted derived images classified into a common group to update data representing said plurality of averaged images, and determines whether updating of data representing said reference images is necessary, wherein, when determining that the updating is necessary, said averaged image data generating unit updates data representing said plurality of reference images in said reference image data storage unit using data representing said plurality of averaged images again and sends a request causing said measured image classifying unit to perform processing for extracting the derived images, and when determining that the updating is not necessary, said averaged image data generating unit provides the generated data representing said plurality of averaged images to said structure determining unit;
• and
• said structure determining unit acquires data representing said plurality of averaged images generated by said averaged image data generating unit and determines a structure of said membrane protein based on the acquired data.

A structure determination method for classifying data representing a plurality of measured images relating to a certain subject of analysis into a plurality of images and for determining a structure of said subject of analysis based on the classification result, said method comprising the steps of:

• generating data representing a plurality of groups of derived images being derived from said measured images by changing a relative position of each of said measured images with respect to said reference images, evaluating similarity between said derived images and said reference images for each of said plurality of groups of derived images, and extracting, from each of said groups of derived images, a plurality of derived images highly similar to any one of said plurality of reference images;
• classifying said plurality of derived images extracted in said step of extracting a plurality of derived images into a plurality of groups, averaging the derived images classified into a common group to generate data representing a plurality of averaged images;
• determining whether updating of data representing said reference images is necessary;
• when it is determined in said step of determining that the updating is necessary, performing the updating of data representing said plurality of reference images using new data representing said plurality of averaged images, and performing the preceding steps starting from said step of generating data representing a plurality of groups of derived images again;
• and
• when it is determined that the updating is not necessary, determining said structure of said subject of analysis based on data representing said plurality of averaged images.

A non-transitory computer readable medium storing a program which causes a computer to function as a structure determination system which classifies data representing a plurality of measured images relating to a certain subject of analysis into a plurality of images and determines a structure of said subject of analysis based on the classification result, said structure determination system comprising:

• a reference image data storage unit which stores data representing a plurality of reference images;
• a measured image classifying unit which generates data representing a plurality of groups of derived images being derived from said measured images by changing a relative position of each of said measured images with respect to said reference images, evaluates similarity between said derived images and said reference images for each of said plurality of groups of derived images, and extracts, from each of said groups of derived images, a plurality of derived images highly similar to any one of said plurality of reference images;
• an averaged image data generating unit which classifies said plurality of derived images extracted by said measured image classifying unit into a plurality of groups, averages the derived images classified into a common group to generate data representing a plurality of averaged images, and determines whether updating of data representing said reference images is necessary, said averaged image data generating unit performing the updating of data representing said plurality of reference images in said reference image data storage unit using data representing said plurality of averaged images and sending a request causing said measured image classifying unit to perform processing for extracting said derived images when determining that the updating is necessary, and said averaged image data generating unit setting the generated data representing a plurality of averaged images as data to be used for structure determination when determining that the updating is not necessary;
• and
• a structure determining unit which acquires data representing said plurality of averaged images generated by said averaged image data generating unit and determines a structure of said subject of analysis based on the acquired data.