Progressive fitting method for protein structure and cryoelectron microscope density map

Abstract

A progressive fitting method for a protein structure and a cryoelectron microscope density map comprises the following steps: firstly, inputting the protein structure and the cryoelectron microscope density map, carrying out trilinear interpolation on the density map, and carrying out normalization processing on voxel values; then, exhaustion of a rotation angle and a translation vector of the protein three-dimensional structure is carried out through Situs software, and the lowest score of the density correlation coefficient is calculated; if the lowest score is higher than a set threshold value, interpolation is conducted on the density map again to change the voxel size, on the basis of the final initial position and direction obtained in the last step, the rotation angle and the translation vector with the lowest score are searched for through the DE algorithm, and finally the best-fitting protein three-dimensional structure is obtained. The progressive fitting method for the protein structure and the cryoelectron microscope density map is high in precision.

Description

Technical Field

[0001] The present invention belongs to the fields of bioinformatics, intelligent optimization and computer application, and in particular relates to a method for progressive fitting of protein structure and cryo-electron microscopy density map. Background Art

[0002] Cryo-EM is a high-resolution electron microscopy technique that uses cryo-EM to image biological samples. It is mainly used in the structural study of biological macromolecules, especially in the fields of biochemistry and molecular biology. With the continuous improvement of imaging resolution, cryo-EM has become an important tool for studying the structure of biological molecules, providing strong support for scientists to unravel the mysteries of life. Refined modeling based on cryo-EM density maps usually includes three steps: atomic structure prediction, fitting it to the density map, and atomic refinement iteration. Among them, fitting the structure of the protein to the cryo-EM density map is an important step in reducing the conformational search space in the subsequent atomic refinement step and ensuring that the density map accurately guides the refinement. Poor initial positions of the predicted structure and the density map will lead to inefficient fitting. For medium and low resolution density maps, the lack of details may make the fitting process more difficult, and advanced computational techniques are needed to fill in the erroneous information in the cryo-EM density map in order to obtain high-resolution protein structures.

[0003] Currently, many algorithms have been developed to automatically search for all possible relative rotations and translations, including: EMFIT, 3SOM, MultiFit, UCSF Chimera, Colores of Situs, ADP_EM. However, although these methods have achieved many successes, the accuracy of existing algorithms is limited by factors such as low resolution of density maps and misalignment between protein structures and density maps; secondly, since the search space is large during the fitting process, exhaustive search is a time-consuming process.

[0004] Therefore, the existing protein structure and cryo-EM density map fitting methods have defects in search algorithms and accuracy and need to be improved. Summary of the invention

[0005] In order to overcome the shortcomings of existing fitting methods in terms of accuracy and search algorithms, the present invention provides a method for asymptotically fitting protein structures and cryo-electron microscopy density maps with higher accuracy.

[0006] The technical solution adopted by the present invention to solve its technical problem is:

[0007] A method for asymptotically fitting a protein structure and a cryo-electron microscopy density map, the method comprising the following steps: 1) inputting a three-dimensional protein structure to be fitted and its corresponding cryo-electron microscopy density map;

[0008] 2) Preprocessing of cryo-EM density maps;

[0009] 3) Use Situs (https: / / situs.biomachina.org / ) to find all the initial positions and directions of the protein three-dimensional structure in the cryo-EM density map and calculate the density correlation coefficient. The calculation formula is:

[0010]

[0011] Where N is the number of voxels in the density map after removing values ​​below the set threshold, ρ E (v i ) is the i-th voxel v in the density map i The density value of Represents the average density value of the density map, ρ M (v i ) is the i-th voxel v of the density map generated based on the three-dimensional structure i The density value of The average density value of the density map generated according to the three-dimensional structure, ρ M (v i ) is calculated as follows:

[0012]

[0013] Among them, x j is the C of the jth residue in the structure a Atomic coordinates, m represents C α The mass of the atom, R represents the resolution of the cryo-EM density map, and v i Represents the position coordinates of the i-th voxel;

[0014] 4) Determine the best fitting structure generated in step 3) according to the density correlation coefficient D. If the density correlation coefficient of the structure is lower than the set threshold, it is used as the best fitting structure; if the density correlation coefficient is higher than the set threshold, a certain number of initial poses with the lowest density correlation coefficient are selected for further optimization;

[0015] 5) The density map is processed again, and then the Differential Evolution (DE) algorithm is used to search for the rotation angle and translation vector with the lowest initial pose score. The optimal position of the structure in the density map is calculated based on the translation vector and rotation angle, and finally the fitted protein structure is obtained.

[0016] Further, the process of step 2) is as follows:

[0017] 2.1) Perform trilinear interpolation on the density map to convert the voxel size in the density map to

[0018] 2.2) Set the voxel threshold v cut , and normalize the voxel values ​​of the density map, and set the voxel values ​​less than the set voxel threshold v cut The density map voxel values ​​are set to zero.

[0019] Further, the process of step 5) is as follows:

[0020] 5.1) Perform trilinear interpolation on the original density map to convert the voxel size in the density map into

[0021] 5.2) Using the initial positions selected in step 4) as the initial rotation angle and translation vector (initial solution) of the DE algorithm to find the best fitting position;

[0022] 5.3) Compare the density correlation coefficients of the best fitting position in step 5.2) and the best fitting structure in step 4), and select the structure corresponding to the smallest density correlation coefficient as the final fitted protein three-dimensional structure.

[0023] The technical concept of the present invention is as follows: first, the protein structure and cryo-electron microscopy density map are input, the density map is trilinearly interpolated, and the voxel value is normalized; then, the rotation angle and translation vector of the protein three-dimensional structure are exhaustively enumerated through the Situs software, and the minimum score of the density correlation coefficient is calculated; if the minimum score is higher than the set threshold, the density map is interpolated again to change the voxel size, and based on the final initial position and direction obtained in the previous step, the DE algorithm is used to search for the rotation angle and translation vector with the lowest score, and finally the best fitting protein three-dimensional structure is obtained.

[0024] The beneficial effects of the present invention are as follows: firstly, a global search is performed for the approximate position of the protein structure in the density map, and then a local search is performed for the precise fitting position and angle of the protein, thereby improving the structure fitting accuracy through a progressive search from global to local. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a basic flow chart of the method for asymptotic fitting of protein structures to cryo-EM density maps.

[0026] Figure 2 This is the cryo-electron microscopy density map corresponding to protein 6gzv.

[0027] Figure 3 This is the final three-dimensional structure of protein 6GZV ​​fitted into the density map. DETAILED DESCRIPTION

[0028] The present invention will be further described below in conjunction with the accompanying drawings.

[0029] Reference Figure 1 to Figure 3, a method for asymptotically fitting a protein structure to a cryo-electron microscopy density map, comprising the following steps:

[0030] 1) Input the protein 3D structure to be fitted and its corresponding cryo-EM density map;

[0031] 2) Preprocess the cryo-EM density map as follows:

[0032] 2.1) Perform trilinear interpolation on the density map to convert the voxel size in the density map to

[0033] 2.2) Set the voxel threshold v cut , and normalize the voxel values ​​of the density map, and set the voxel values ​​less than the set voxel threshold v cut The density map voxel values ​​are set to zero;

[0034] 3) Use Situs (https: / / situs.biomachina.org / ) to find all the initial positions and directions of the protein three-dimensional structure in the cryo-EM density map and calculate the density correlation coefficient. The calculation formula is:

[0035]

[0036] Where N is the number of voxels in the density map after removing values ​​below the set threshold, ρ E (v i ) is the i-th voxel v in the density map i The density value of Represents the average density value of the density map, ρ M (v i ) is the i-th voxel v of the density map generated based on the three-dimensional structure i The density value of The average density value of the density map generated according to the three-dimensional structure, ρ M (v i ) is calculated as follows:

[0037]

[0038] Among them, x j is the C of the jth residue in the structure α Atomic coordinates, m represents C α The mass of the atom, R represents the resolution of the cryo-EM density map, and v i Represents the position coordinates of the i-th voxel;

[0039] 4) Determine the best fitting structure generated in step 3) according to the density correlation coefficient D. If the density correlation coefficient of the structure is lower than the set threshold, it is used as the best fitting structure; if the density correlation coefficient is higher than the set threshold, a certain number of initial poses with the lowest density correlation coefficient are selected for further optimization;

[0040] 5) The density map is processed again; then the Differential Evolution (DE) algorithm is used to search for the rotation angle and translation vector with the lowest initial pose score, and the optimal position of the structure in the density map is calculated based on the translation vector and rotation angle, and finally the fitted protein structure is obtained. The process is as follows;

[0041] 5.1) Perform trilinear interpolation on the original density map to convert the voxel size in the density map into

[0042] 5.2) Using the 20 initial poses selected in step 4) as the initial rotation angle and translation vector of the DE algorithm to find the best fitting position;

[0043] 5.3) Compare the density correlation coefficients of the best fitting position in step 5.2) and the best fitting structure in step 4), and select the structure corresponding to the smallest density correlation coefficient as the final fitted protein three-dimensional structure.

[0044] This example uses protein 6gzv as an example, and a protein complex cryo-electron microscopy fitting method based on a search algorithm includes the following steps:

[0045] 1) Input the protein 3D structure to be fitted and its corresponding cryo-EM density map;

[0046] 2) Preprocess the cryo-EM density map as follows:

[0047] 2.1) Perform trilinear interpolation on the density map to convert the voxel size in the density map to

[0048] 2.2) Set the voxel threshold to 0.005 and normalize the density map voxel values, setting the density map voxel values ​​that are less than the set voxel threshold of 0.005 to zero;

[0049] 3) Use Situs (https: / / situs.biomachina.org / ) to find all the initial positions and directions of the protein three-dimensional structure in the cryo-EM density map and calculate the density correlation coefficient. The calculation formula is:

[0050]

[0051] Where N is the number of voxels in the density map after removing values ​​below the set threshold, ρ E (v i ) is the i-th voxel v in the density map i The density value of Represents the average density value of the density map, ρ M (v i ) is the i-th voxel v of the density map generated based on the three-dimensional structure i The density value of The average density value of the density map generated according to the three-dimensional structure, ρ M (v i ) is calculated as follows:

[0052]

[0053] Among them, x j is the C of the jth residue in the structure α Atomic coordinates, m represents C α The mass of the atom, R represents the resolution of the cryo-EM density map, and v i Represents the position coordinates of the i-th voxel;

[0054] 4) Determine the best fitting structure generated in step 3) according to the density correlation coefficient D. If the density correlation coefficient of the structure is lower than 0.4, it is used as the best fitting structure; if the density correlation coefficient is higher than 0.4, select the 20 initial poses with the lowest density correlation coefficient for further optimization;

[0055] 5) The density map is processed again; then the Differential Evolution (DE) algorithm is used to search for the rotation angle and translation vector with the lowest initial pose score, and the optimal position of the structure in the density map is calculated based on the translation vector and rotation angle, and finally the fitted protein structure is obtained. The process is as follows:

[0056] 5.1) Perform trilinear interpolation on the original density map to convert the voxel size in the density map into

[0057] 5.2) Using the 20 initial poses selected in step 4) as the initial rotation angle and translation vector of the DE algorithm to find the best fitting position;

[0058] 5.3) Compare the density correlation coefficients of the best fitting position in step 5.2) and the best fitting structure in step 4), and select the structure corresponding to the smallest density correlation coefficient as the final fitted protein three-dimensional structure.

[0059] In this example, protein 6gzv is used as an example to obtain the fitted structure of the protein using the above method. The predicted results are as follows: Figure 3As shown, the root mean square deviation between it and the protein structure library is

[0060] The above description is about the excellent effect shown by an embodiment of the present invention. Obviously, the present invention is not only suitable for the above embodiment, but also can be implemented with various changes without departing from the basic spirit of the present invention and without exceeding the content involved in the essential content of the present invention.

Claims

1. A method for asymptotically fitting protein structures to cryo-electron microscopy density maps, characterized in that: The method comprises the following steps: 1) Input the protein 3D structure to be fitted and its corresponding cryo-EM density map; 2) Preprocessing of cryo-EM density maps; 3) Use Situs to find all the initial positions and directions of the protein three-dimensional structure in the cryo-EM density map and calculate the density correlation coefficient. The calculation formula is: Where N is the number of voxels in the density map after removing values ​​below the set threshold, ρ E (v i ) is the i-th voxel v in the density map i The density value of Represents the average density value of the density map, ρ M (v i ) is the i-th voxel v of the density map generated based on the three-dimensional structure i The density value of The average density value of the density map generated according to the three-dimensional structure, ρ M (v i ) is calculated as follows: Among them, x j is the C of the jth residue in the structure a Atomic coordinates, m represents C a The mass of the atom, R represents the resolution of the cryo-EM density map, and v i Represents the position coordinates of the i-th voxel; 4) Determine the best fitting structure generated in step 3) according to the density correlation coefficient D. If the density correlation coefficient of the structure is lower than the set threshold, it is used as the best fitting structure; if the density correlation coefficient is higher than the set threshold, select the 20 initial poses with the lowest density correlation coefficient for further optimization; 5) The density map is processed again; then the Differential Evolution algorithm is used to search for the rotation angle and translation vector with the lowest initial pose score, and the optimal position of the structure in the density map is calculated based on the translation vector and rotation angle, and finally the fitted protein structure is obtained.

2. A method for asymptotically fitting a protein structure and a cryo-electron microscopy density map according to claim 1, characterized in that: The process of step 2) is as follows; 2.1) Perform trilinear interpolation on the density map to convert the voxel size in the density map to 2.2) Set the voxel threshold to 0.005 and normalize the density map voxel values, setting the density map voxel values ​​that are less than the set voxel threshold of 0.005 to zero.

3. A method for asymptotically fitting a protein structure and a cryo-electron microscopy density map according to claim 1 or 2, characterized in that: The process of step 5) is as follows; 5.1) Perform trilinear interpolation on the original density map to convert the voxel size in the density map into 5.2) Using the 20 initial poses selected in step 4) as the initial rotation angle and translation vector of the DE algorithm to find the best fitting position; 5.3) Compare the density correlation coefficients of the best fitting position in step 5.2) and the best fitting structure in step 4), and select the structure corresponding to the smallest density correlation coefficient as the final fitted protein three-dimensional structure.

Images