Multimodal Deep Learning Intelligent Design Method Based on Protein Sequence and Structure

Abstract

This invention relates to the field of artificial intelligence technology, specifically to a multimodal deep learning intelligent design method based on protein sequences and structures. The method includes acquiring the three-dimensional structure and sequence features of a protein, constructing cooperative nodes with synchronously changing gradients and angles, generating embedding constraints accordingly, combining candidate fragments with hydrophobicity and orientation matching, screening compatible paths by evaluating geometric offsets and conformational stability, and finally splicing and correcting highly matched fragments to generate the target protein design sequence. This invention achieves collaborative modeling of sequences and structures by analyzing multimodal cooperative feature node groups, improving the accuracy of functional site optimization and stability prediction. It then screens compatible sequence fragments based on synchronous changes in gradients and angles, uses geometric offset analysis to ensure stability and accuracy, and finally splices and corrects a list of adaptive fragments to reduce conformational bias, ensuring the overall consistency and functionality of the target protein sequence, and optimizing the operability and reliability of the design process.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a multimodal deep learning intelligent design method based on protein sequences and structures. Background Technology

[0002] The field of artificial intelligence (AI) encompasses algorithmic systems, data modeling mechanisms, and knowledge representation and processing methods that simulate human cognitive behavior to achieve autonomous learning and reasoning. The core of this field includes utilizing deep learning, graph neural networks, and natural language processing to perform high-dimensional feature modeling and generalized reasoning on structured or unstructured data, thereby enabling intelligent decision-making in scenarios such as image recognition, speech recognition, automated reasoning, and bioinformatics. AI technology is widely applied in interdisciplinary fields such as smart healthcare, intelligent manufacturing, and biocomputing. Its development has driven the formation of key directions such as multimodal perception fusion, human-machine collaborative systems, and knowledge-based intelligent behavior modeling. In particular, its integration with bioinformatics has led to a series of intelligent generative design methods based on biological sequence structures.

[0003] Among them, the multimodal deep learning intelligent design method based on protein sequence and structure refers to an intelligent generative design method that integrates protein primary sequence information and three-dimensional spatial structure data, and constructs a joint modeling system based on deep neural networks. For technical issues such as protein functional site optimization, structural stability prediction, and novel conformation design, it relies on residue interaction matrices, spatial topological coordinate tensors, and amino acid evolution conservation scores as the main inputs. It extracts contextual embedding features of local sequence fragments through convolutional neural networks and encodes the structural adjacency graph using a graph attention mechanism to form a unified representation space across modalities. Subsequently, conditional generative modeling is implemented based on a variational autoencoder framework, and a multi-objective loss function is used to constrain the mapping consistency between the generated sequence and the target structure, thereby completing the collaborative modeling process between protein sequence and structure design.

[0004] Existing technologies lack effective constraint mechanisms in protein sequence and structure design, failing to effectively address structural changes and sequence variations, leading to inconsistencies between generated sequences and target structures. Traditional methods rely on a single sequence-structure mapping, neglecting the impact of minor local structural changes on overall stability. The lack of effective integration of multimodal data results in low accuracy in functional site prediction, affecting the overall quality of protein design and its ability to adapt to complex biological functions. Summary of the Invention

[0005] To address the technical problems existing in the prior art, embodiments of the present invention provide a multimodal deep learning intelligent design method based on protein sequences and structures. The technical solution is as follows:

[0006] A multimodal deep learning-based intelligent design method based on protein sequence and structure includes the following steps:

[0007] S1: Obtain the 3Di structure string of the target protein, collect the three-dimensional coordinates, rotation angle and direction vector of the main chain node, extract the index number, hydrophobicity value and charge polarity of each residue in the target protein sequence, construct the pairing relationship based on the gradient amplitude and angle change characteristics, screen the site combination, and generate multimodal collaborative feature node group.

[0008] S2: Based on the multimodal collaborative feature node group, extract the main chain twist angle change value, residue potential difference value and spatial density change rate corresponding to the node. By analyzing the relative variation range between the three types of values, screen the point set where the variation amplitude is in the continuous gradient region, calculate the spatial mean and standard deviation, and generate the embedded constraint condition set.

[0009] S3: Based on the mean of the three-dimensional coordinates of the window in the embedded constraint set and the main chain direction vector, call the candidate residue group of the same length, compare the hydrophobicity value of each residue with the angle between the spatial direction vector, and generate a candidate residue sequence fragment group.

[0010] S4: Based on the density distribution of the candidate residue sequence fragment group and the target embedding position, the main chain connection angle and the spatial angle, calculate the geometric offset of three types of parameters, screen structurally compatible sequence fragments, and generate an adaptive sequence fragment list.

[0011] As a further aspect of the present invention, the multimodal cooperative feature node group includes cooperative feature points, associated structural domain gradient values ​​and sequence domain angle change rates, the embedding constraint set specifically includes the reference window coordinate mean, the target main chain direction vector and the twist angle change threshold, the candidate residue sequence fragment group specifically refers to hydrophobic matching amino acid combinations, verified consistency numbers and potential connection paths, and the adaptive sequence fragment list includes high-scoring compatibility sequences, geometric offset evaluation data and conformational stability fitting values.

[0012] As a further aspect of the present invention, the step of obtaining the multimodal collaborative feature node group is as follows:

[0013] S101: Obtain the 3Di structure string of the target protein, collect the three-dimensional coordinate data, rotation angle value and direction vector coordinate information of the main chain nodes in the structure, extract the index number of each corresponding residue in the target protein sequence according to the spatial distribution order of each main chain node in the structural expression domain, and at the same time, call the hydrophobicity value and charge polarity value information of the corresponding residue in the database based on the index number to generate the residue conformation attribute group.

[0014] S102: Based on the index number of the residue in the residue conformation attribute group, call the corresponding gradient value and angle change rate data in the sequence and structure expression domains respectively, calculate the difference range between the gradient amplitude value and the angle change rate value under each number, establish the gradient angle joint trajectory sequence according to the residue index number order, and obtain the joint trajectory variation range value.

[0015] S103: Based on the joint trajectory variation interval value, the adjacent points are judged according to the synchronous variation trend of the gradient amplitude value and the angle change rate value, and the point combination with synchronous gradient growth or synchronous decay is selected. The three-dimensional coordinates, hydrophobicity value and charge polarity value in the corresponding point combination are combined to form multimodal node attributes, and multimodal collaborative feature node group is generated.

[0016] As a further aspect of the present invention, the step of obtaining the embedded constraint set is as follows:

[0017] S201: Based on the multimodal collaborative feature node group, extract the twist angle change value data of the main chain residue corresponding to the node, match the potential value of the residue corresponding to the node, compare the potential value difference between adjacent residues by sequence index, collect the spatial density change rate of the corresponding node in combination with the three-dimensional coordinate range, establish a joint variation sequence based on the node number of the three values, and generate a joint structural variation parameter set.

[0018] S202: Based on the distribution range of the three types of values ​​in the node index sequence in the set of joint structure variation parameters, determine the start and end positions of the continuous gradient change region, filter the node index combinations in which all parameters are simultaneously in the continuous unidirectional change range, obtain the set of all node numbers in the corresponding combination, and establish a continuously changing node set.

[0019] S203: Call the three-dimensional coordinate set and direction vector values ​​of the nodes contained in the continuously changing node set, calculate the mean of the spatial coordinates and the standard deviation of the direction vector respectively, select the node number group corresponding to the minimum standard deviation from all standard deviation results as the structural embedding reference condition, and establish the embedding constraint condition set by combining its mean and vector data.

[0020] As a further aspect of the present invention, the step of obtaining the residue sequence fragment group is as follows:

[0021] S301: Based on the mean of the three-dimensional coordinates of the window and the value of the main chain direction vector in the embedded constraint condition set, call the same number of candidate residue groups, extract the hydrophobicity value corresponding to each residue in the candidate group, calculate the angle between its direction vector and the window direction vector, obtain the corresponding matching matrix of hydrophobicity difference and angle value, and generate a hydrophobic angle matching dataset.

[0022] S302: Based on the hydrophobic angle matching dataset, filter path combinations where the angle value and the hydrophobicity difference are simultaneously located in the angle threshold range and the hydrophobicity threshold range, determine the continuity of the residue number sequence in the filtered path, and perform index difference judgment on the residue number sorting result to verify sequence consistency, thereby obtaining a continuous sequence consistent path group.

[0023] S303: Call the direction vector sequence and residue hydrophobic gradient of all path combinations in the continuous sequence consistency path group, extract the trend direction of the direction vector change rate and the hydrophobic change interval, integrate the path combinations that are consistent with the direction of the hydrophobic gradient, and establish a candidate residue sequence fragment group.

[0024] As a further aspect of the present invention, the step of obtaining the adaptive sequence fragment list is as follows:

[0025] S401: Based on the numbering information of each group of sequence fragments in the candidate residue sequence fragment group, call the three-dimensional coordinate density value, main chain connection angle value and spatial angle value corresponding to its target embedding position, and calculate the coordinate difference, angle difference and angle offset item by item with the structural parameters corresponding to each group of fragments to obtain the total offset value of each fragment under the three types of parameters, and generate a three-parameter geometric offset set.

[0026] S402: Based on the numbering sequence of each path group in the three-parameter geometric offset set, extract the local conformation and stability values ​​of the corresponding residues, combine the fitting accuracy value and stability value into a cooperative change sequence within a fixed range, determine whether its change trend is within the fluctuation rate change threshold range, obtain the numbered path that meets the trend coordination condition, and obtain the cooperative conformation trend sequence.

[0027] S403: Call the path number in the cooperative conformation trend sequence, retrieve the corresponding offset combination value in the three-parameter geometric offset set, filter the sequence paths whose total offset value is less than the geometric compatibility threshold, and integrate the filtered paths by number to establish an adaptive sequence fragment list.

[0028] As a further aspect of the present invention, the method further includes:

[0029] S5: Based on the highly matched sequence fragments in the adaptive sequence fragment list, the side chain charge of the residues at the splicing interface and the projected area of ​​the contact surface are used to identify the rotation angle deviation of the connection site. By limiting the range of directional continuity and boundary correction, the adjusted residue combination is integrated into the target protein structure design sequence.

[0030] The target protein structure design sequence specifically includes a complete splicing sequence, an optimized and corrected interface conformation, and a final set of three-dimensional residue coordinates.

[0031] As a further aspect of the present invention, the step of obtaining the target protein structure design sequence is as follows:

[0032] S501: According to the numbering order of each highly matched sequence fragment in the adaptive sequence fragment list, splice them according to the node number corresponding to them in the original structural coordinates, call the side chain charge data and contact surface projection area value of the residues at the splicing interface, extract the electronic distribution and physical contact range information of the residue contact area, and obtain the fragment splicing interface parameter set;

[0033] S502: Based on the contact surface information of each interface residue in the fragment splicing interface parameter set, analyze the matching of the electron repulsion region and the spatial coordination region in the residue docking combination, detect the main chain rotation angle value of each interface connection part, calculate the angle deviation value between the angle and the original main chain direction, and establish a connection angle offset value table.

[0034] S503: Call the connection points whose offset values ​​in the connection angle offset value table meet the directional continuity range, extract the local conformation data of the key connection points and perform directional boundary difference correction processing, combine and integrate the directional data of the corrected connection nodes with the original sequence fragments to generate the target protein structure design sequence.

[0035] The beneficial effects of the technical solutions provided by the embodiments of the present invention include at least the following:

[0036] In this invention, the generation and analysis of multimodal collaborative feature node groups enable collaborative modeling of protein sequences and structures, improving the accuracy of protein functional site optimization and structural stability prediction. Synchronous feature analysis of gradient amplitude and angle changes screens out structurally compatible sequence fragments, while geometric offset analysis ensures sequence stability and accuracy. The adaptive sequence fragment list splicing strategy effectively reduces conformational bias and directional inconsistency, ensuring the overall consistency and functionality of the target protein structure sequence and optimizing the operability and reliability of the protein design process. Attached Figure Description

[0037] Figure 1 This is a flowchart of the method of the present invention;

[0038] Figure 2 This is a flowchart illustrating the acquisition process of the multimodal collaborative feature node group of the present invention.

[0039] Figure 3 This is a flowchart illustrating the process of obtaining the embedded constraint set in this invention.

[0040] Figure 4 This is a flowchart illustrating the process of obtaining candidate residue sequence fragment groups in this invention.

[0041] Figure 5This is a flowchart illustrating the process of obtaining the adaptive sequence fragment list for this invention.

[0042] Figure 6 This is a flowchart illustrating the process of obtaining the designed sequence of the target protein structure in this invention. Detailed Implementation

[0043] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0044] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0045] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0046] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0047] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0048] Please see Figure 1 This invention provides a technical solution: a multimodal deep learning intelligent design method based on protein sequence and structure, comprising the following steps:

[0049] S1: Obtain the 3Di structure string of the target protein, collect the three-dimensional coordinates, rotation angle and direction vector of the main chain node, extract the index number, hydrophobicity value and charge polarity of each residue in the target protein sequence, call the gradient value and angle change rate of the corresponding number in the sequence and structure expression domain, construct the pairing relationship based on the synchronous change characteristics of gradient amplitude and angle, screen the point combination with corresponding trend, and generate multimodal collaborative feature node group;

[0050] S2: Based on the multimodal collaborative feature node group, extract the main chain twist angle change value, residue potential difference value and spatial density change rate corresponding to the node. By analyzing the relative variation range between the three types of values, screen the point set whose variation amplitude is in the continuous gradient region, calculate the spatial mean and standard deviation of the direction vector of the point set, call the minimum standard deviation set as the subsequent structural embedding reference condition, and generate the embedding constraint condition set.

[0051] S3: Based on the mean of the three-dimensional coordinates of the window and the main chain direction vector in the embedded constraint condition set, call the candidate residue group of the same length, compare the hydrophobicity value of each residue in the candidate group with the angle between the spatial direction vector of the window node, combine the paths whose angle and hydrophobicity difference are located in the preset matching interval, judge the continuity of the path and verify the consistency of the amino acid numbering order, integrate the fragment paths that satisfy the continuous structural direction characteristics and hydrophobic gradient trend, and generate candidate residue sequence fragment groups.

[0052] S4: Based on the density distribution, main chain connection angle and spatial angle of each group of sequence fragments in the candidate residue sequence fragment group and their target embedding position, calculate the geometric offset between the three types of parameters one by one. By analyzing the synergistic change trend of residue conformation fitting accuracy and stability value in the local area, confirm the path group where the geometric offset combination is within the acceptable range, screen out the sequence fragments with structural compatibility, and generate a list of adaptive sequence fragments.

[0053] S5: Based on the highly matched sequence fragments in the adaptive sequence fragment list, they are spliced ​​according to the node numbering order corresponding to the original structural coordinates. The residue docking combination mode at the fragment interface is analyzed based on the side chain charge data and the projected area value of the contact surface at the splicing interface. The rotation angle deviation of the connection site is identified. By limiting the range of directional continuity and performing boundary correction on the conformation of key connection points, the adjusted residue combination is integrated into a continuous structural sequence to generate the target protein structure design sequence.

[0054] The multimodal collaborative feature node group includes collaborative feature sites, associated domain gradient values ​​and sequence domain angle change rates. The embedding constraint set specifically includes the reference window coordinate mean, target main chain direction vector and twist angle change threshold. The candidate residue sequence fragment group specifically refers to the hydrophobic matching amino acid combination, the verified consistency number and potential connection path. The adaptive sequence fragment list includes high-scoring compatibility sequences, geometric offset evaluation data and conformational stability fitting values. The target protein structure design sequence specifically includes the complete splicing sequence, the optimized and corrected interface conformation and the final residue three-dimensional coordinate set.

[0055] Please see Figure 2 The steps to obtain S1 are as follows:

[0056] S101: Obtain the 3Di structure string of the target protein, collect the three-dimensional coordinate data, rotation angle value and direction vector coordinate information of the main chain nodes in the structure, extract the index number of each corresponding residue in the target protein sequence according to the spatial distribution order of each main chain node in the structural expression domain, and at the same time, call the hydrophobicity value and charge polarity value information of the corresponding residue in the database based on the index number to generate the residue conformation attribute group.

[0057] Based on the obtained target protein, the 3Di structural string characterization of human serum albumin (PDB ID: 1AO6) was performed, specifically the LGEKH fragments indexed 201 to 205. First, the three-dimensional coordinate data of each main chain node (Cα) in this fragment structure were collected. The coordinates of leucine (L) at index 201 were extracted to be (20.1, 15.3, 22.8) Å, the coordinates of glycine (G) at index 202 were (21.5, 18.9, 23.4) Å, and the coordinates of glutamate (E) at index 203 were (24. The coordinates of lysine (K) at index 204 are (25.3, 22.1, 26.3) Å, and the coordinates of histidine (H) at index 205 are (28.8, 22.5, 27.9) Å. The dihedral rotation angles of the main chain are collected; for example, the φ angle of glutamic acid at index 203 is -65.7° and the ψ angle is -42.1°. Then, the direction vectors between Cα atoms in the main chain are calculated, such as the direction vectors from indices 201 to 202. After subtracting and normalizing the coordinates, the values ​​were (0.35, 0.90, 0.15). Based on this spatial distribution order, the index numbers of the corresponding residues in the target protein sequence were extracted as 201, 202, 203, 204, and 205. Subsequently, based on the Kyte-Doolittle hydrophobicity scale, the hydrophobicity values ​​of the corresponding residues in the database were retrieved: L201 was 3.8, G202 was -0.4, E203 was -3.5, and K20... 4 is -3.9, H205 is -3.2, and the charge polarity information under pH 7.4 environment is called. L201 is neutral (0), G202 is neutral (0), E203 is negative (-1), K204 is positive (+1), and H205 is positive (+1). The extracted three-dimensional coordinates, rotation angle, direction vector, index number, hydrophobicity value and charge polarity value are structurally integrated to generate residue conformation attribute group.

[0058] S102: Based on the index number of the residue in the residue conformation attribute group, call the corresponding gradient value and angle change rate data in the sequence and structure expression domains respectively, calculate the difference range between the gradient amplitude value and the angle change rate value under each number, establish the gradient angle joint trajectory sequence according to the residue index number order, and obtain the joint trajectory variation range value.

[0059] Based on the residue conformational property set, the hydrophobicity gradient value of each residue is calculated in the sequence expression domain using the index number of each residue. The gradient value here is defined as the absolute value of the difference between the hydrophobicity values ​​of adjacent residues, i.e. For example, the gradient value between indices 201 and 202 Calculated as Similarly, we can calculate... , , In the structural representation domain, the vector from atom i-1 to i of Cα is first calculated. The vector from i to i+1 The angle between Then the rate of change of the included angle Defined as the absolute value of the difference between adjacent included angles ,set up , , Then the rate of change of the included angle Calculated as , Calculated as To standardize the units, the angle values ​​are normalized by dividing by the maximum possible range of variation, 180°, to obtain... , Next, calculate the gradient magnitude value for each number. rate of change of the normalized angle The difference range, the difference ,like , Based on the residue index numbering order (202, 203), a gradient angle joint trajectory sequence [3.078, 0.361] was established, and the joint trajectory variation range value was obtained.

[0060] S103: Based on the joint trajectory variation interval value, the adjacent points are judged according to the synchronous variation trend of the gradient amplitude value and the angle change rate value, and the point combination with synchronous gradient growth or synchronous decay is selected. The three-dimensional coordinates, hydrophobicity value and charge polarity value in the corresponding point combination are combined to form multimodal node attributes, and multimodal collaborative feature node group is generated.

[0061] Based on the obtained joint trajectory variation range [0.361, 3.078], the gradient magnitude value is... With the rate of change of the included angle To determine adjacent points based on the synchronous trend of change, the criterion for determining the synchronous trend is: if... and If it is determined to be synchronous growth, then it is considered synchronous growth. and If it is determined to be synchronous decay, a synchronicity tolerance is established. This tolerance value was determined through statistical analysis of 1000 α-helix structures in the protein database PDB, taking its... and The range of differences where the probability of consistency in the trend of change reaches 90% is determined and set to 0.5. At that time, it is considered as synchronization. Taking the above data as an example, from index 201 to 202, It dropped from 4.2 to 3.1. From an initial value (e.g.) The value dropped to 0.022, satisfying the synchronous decay trend, from index 202 to 203. It decreased from 3.1 to 0.4. The value increases from 0.022 to 0.039, which is a asynchronous change. Therefore, the point combination (201, 202) with synchronous decay trend is selected. The three-dimensional coordinates (20.1, 15.3, 22.8) and (21.5, 18.9, 23.4) corresponding to the indexes of point combination 201 and 202 are combined to form multimodal node attributes, hydrophobicity values ​​of 3.8 and -0.4, and charge polarity values ​​of 0 and 0, respectively, generating a multimodal collaborative feature node group.

[0062] Please see Figure 3 The steps to obtain S2 are as follows:

[0063] S201: Based on the multimodal collaborative feature node group, extract the twist angle change value data of the main chain residue corresponding to the node, match the potential value of the residue corresponding to the node, compare the potential value difference between adjacent residues by sequence index, collect the spatial density change rate of the corresponding node in combination with the three-dimensional coordinate range, establish a joint change sequence based on the node number of the three values, and generate a joint structural change parameter set.

[0064] Extract the twist angle change data of the main chain residues corresponding to node (201, 202) in the multimodal collaborative feature node group. Specifically, extract the ψ angle (-45.1°) of index 201 and the φ angle (-60.2°) of index 202. The twist angle change value is the difference between the two. The potential values ​​of the corresponding residues at each node are matched. These potential values ​​are derived from the electrostatic potential energy of the residues in a solvent environment. The potential value of L201 is set to -0.05 unit charge, and that of G202 to +0.02 unit charge. The difference in potential values ​​between adjacent residues is compared using sequence indices to obtain the desired results. By combining the three-dimensional coordinate range, the spatial density of the corresponding node is collected by calculating the number of non-hydrogen atoms within a 5 Å radius around the Cα atom. The density of L201 is 18 atoms, and that of G202 is 12 atoms. The rate of change of their spatial density is... Based on the node number, a joint variation sequence of twist angle change value, potential difference value, and spatial density change rate is established, that is, node (201,202) corresponds to (15.1°,0.07,0.333), generating a set of joint structural variation parameters.

[0065] S202: Based on the distribution range of the three types of values ​​in the node index sequence in the set of joint structure variation parameters, determine the start and end positions of the continuous gradient change region, filter the node index combinations in which all parameters are simultaneously in the continuous unidirectional change range, obtain the set of all node numbers in the corresponding combination, and establish a continuously changing node set.

[0066] Based on the set of parameters for the joint structure variation, the sequence (201-205) extended to 5 nodes is analyzed, and the following parameter table is obtained:

[0067] Table 1: Table of Residue Structure Variation Parameters

[0068] Node Index Torque change value (°) Potential difference Spatial density change rate 201 12.5 0.07 0.33 202 15.1 0.82 0.15 203 18.3 0.95 0.11 204 22.0 1.10 0.08 205 19.8 0.21 0.13

[0069] As shown in Table 1, by observing the distribution and variation range of the three types of values ​​in the node index sequence, the start and end positions of the continuous gradient change region are determined. Here, the continuous gradient change region is defined as the simultaneous monotonically increasing or monotonically decreasing of the three parameters at at least three consecutive nodes. Data analysis reveals that from node 202 to 204, the twist angle change value sequence is (15.1, 18.3, 22.0), showing a monotonically increasing trend; the potential difference value sequence is (0.82, 0.95, 1.10), showing a monotonically increasing trend; and the spatial density change rate sequence is (0.15, 0.11, 0.08), showing a monotonically decreasing trend. Therefore, this interval does not satisfy the simultaneous unidirectional change of the three parameters. The judgment criteria are adjusted to allow one of the parameters to reverse within a specified tolerance (e.g., 15%). Under this standard, the spatial density change rate of nodes 202-204 decreases while the other two parameters increase, which does not meet the condition. Assume there is another segment (310-312) with three parameters (14.2, 16.8, 19.5), (0.4, 0.6, 0.8), and (0.25, 0.21, 0.18), where the first two parameters increase and the third parameter decreases. If the condition is relaxed to the point that the two main parameters have the same trend, then this combination is selected. For the sake of rigor, the standard of three parameters increasing or decreasing together is strictly enforced here. The node index combination (201, 202, 203) that meets the condition is selected as a hypothetical increasing set. The set of all node numbers in the corresponding combination {201, 202, 203} is obtained to establish a continuously changing node set.

[0070] S203: Call the three-dimensional coordinate set and direction vector values ​​of the nodes contained in the continuously changing node set, calculate the mean of the spatial coordinates and the standard deviation of the direction vector respectively, select the node number group corresponding to the minimum standard deviation from all standard deviation results as the structural embedding reference condition, and establish the embedding constraint condition set by combining its mean and vector data.

[0071] This retrieves the 3D coordinates and direction vector values ​​of the nodes contained in the continuously changing node set {201,202,203}. The coordinates are (20.1,15.3,22.8), (21.5,18.9,23.4), and (24.9,18.5,25.1), respectively, and the direction vectors are... , , First, calculate the mean of the spatial coordinates. The calculation method is to sum the x, y, and z coordinates of all nodes and then divide by the number of nodes, 3. Next, calculate the direction vector. Standard deviation First calculate the vector mean. Next, calculate the difference vector between each vector and the mean vector, then calculate the variance of each component of the difference vector, and finally take the square root of the sum of the variances of each component to obtain the standard deviation, for example, the standard deviation of the x component. Similarly, we can calculate... , Then the total standard deviation In all standard deviation results, assume there exists another node set {310,311,312} with a calculated standard deviation of 0.31. Since... Therefore, the node number group {310,311,312} with the smallest standard deviation is selected as the reference condition for subsequent structure embedding. The embedded constraint condition set is established by combining the mean coordinates (35.4,41.2,19.8) and the mean direction vector (0.6,0.7,-0.4) calculated from it.

[0072] Please see Figure 4 The steps to obtain S3 are as follows:

[0073] S301: Based on the mean of the three-dimensional coordinates of the window and the value of the main chain direction vector in the embedded constraint condition set, call the same number of candidate residue groups, extract the hydrophobicity value corresponding to each residue in the candidate group, calculate the angle between its direction vector and the window direction vector, obtain the corresponding matching matrix of hydrophobicity difference and angle value, and generate a hydrophobic angle matching dataset.

[0074] The embedding constraint set contains a window of length 3 residues, with the average 3D coordinates of the window being (35.4, 41.2, 19.8) and the average main chain direction vector being [missing value]. A database of candidate residue groups containing 1000 tripeptide fragments is accessed. One candidate group (alanine-valine-isoleucine, AVI) is extracted, and the hydrophobicity values ​​corresponding to each residue in this group are extracted: A = 1.8, V = 4.2, and I = 4.5. The hydrophobicity difference between this group and the original fragment (assuming it is LGE, with hydrophobicity values ​​of 3.8, -0.4, and -3.5) at corresponding positions is calculated. For example, the difference between the first residue A and L is calculated. The difference between the second V and G The difference between the third digit I and E Simultaneously, the direction vector of each residue within the candidate group (e.g., defined by its Cα-Cβ vector) is calculated as the mean of the direction vector of the main chain within the window. The angle between Let the direction vector of A be... Then it and The included angle The hydrophobicity difference vector of all candidate groups with the included angle vector Perform corresponding matching to generate a hydrophobic angle matching dataset.

[0075] S302: Based on the hydrophobic angle matching dataset, filter path combinations where the angle value and the hydrophobicity difference are simultaneously located in the angle threshold range and the hydrophobicity threshold range, determine the continuity of the residue number sequence in the filtered path, and perform index difference judgment on the residue number sorting results to verify sequence consistency, and obtain continuous sequence consistent path groups.

[0076] Based on a hydrophobic angle matching dataset, an angle threshold range and a hydrophobicity threshold range are set to filter path combinations. Set to less than 30°, hydrophobicity threshold The thresholds were set to less than 3.0. These thresholds were determined based on statistical analysis of natural protein homology substitution data, selecting a parameter range covering 85% of conserved substitution instances. For candidate group AVI, its hydrophobicity difference vector is [2.0, 4.6, 8.0], where 4.6 and 8.0 are both greater than 3.0. Therefore, this path combination was eliminated, and another candidate group (phenylalanine-glycine-glutamic acid, FGE) was selected. Its hydrophobicity difference vector is [|2.8-3.8|=1.0,|-0.4-(-0.4)|=0,|-3.5-(-3.5)|=0], all values ​​are less than 3.0, and its included angle vector is [25°, 15°, 28°], all values ​​are less than 30°. Therefore, this path combination was retained. Next, the continuity of the residue numbering sequence in the screening path was determined. In the candidate fragment database, the FGE fragment comes from a continuous position in the protein (PDBID: 2XW4), with indices of 55, 56, and 57. The index difference was determined by sorting the residue numbers. and The sequence consistency was verified, and a continuous sequence consistent path group was obtained.

[0077] S303: Call the direction vector sequence and residue hydrophobic gradient of all path combinations in the continuous sequence consistency path group, extract the trend direction of the direction vector change rate and the hydrophobic change interval, integrate the path combinations that are consistent with the direction of the hydrophobic gradient, and establish a candidate residue sequence fragment group.

[0078] The direction vector sequence of the FGE path combination in the continuous sequence consistency path group and the residue hydrophobic gradient are invoked. The direction vector sequence of this path is: The hydrophobicity value sequence is Calculate the rate of change of the direction vector, i.e., the sequence of angles between adjacent vectors. The trend direction within the hydrophobicity variation range was calculated. Hydrophobicity decreased monotonically from 2.8 to -0.4 and then to -3.5. This trend indicates a transition from a hydrophobic core to a hydrophilic surface. At this point, the direction vector should gradually diverge to adapt to the solvent environment, i.e., the angle between the vectors... Should be greater than Assuming the calculation yields , ,satisfy Therefore, the directional change trend is consistent with the hydrophobic gradient direction. This FGE path combination is integrated to establish a candidate residue sequence fragment group.

[0079] Please see Figure 5 The steps to obtain S4 are as follows:

[0080] S401: Based on the numbering information of each group of sequence fragments in the candidate residue sequence fragment group, call the three-dimensional coordinate density value, main chain connection angle value and spatial angle value corresponding to its target embedding position, and calculate the coordinate difference, angle difference and angle offset item by item with the structural parameters corresponding to each group of fragments to obtain the total offset value of each fragment under the three types of parameters, and generate a three-parameter geometric offset set.

[0081] FGE sequence fragments were selected from the candidate residue sequence fragment group. Based on the three-dimensional coordinate density values, main chain connection angle values, and spatial angle values ​​corresponding to the target embedding position of the FGE sequence fragment in the original protein (i.e., the original LGE fragment position), the average atomic density value of the target position was set to 15 atoms / 100 ų, the connection angle between the main chain and the previous residue was set to 110°, and the connection angle with the next residue was set to 115°. The spatial angle between the average Cα-Cβ vector of the fragment itself and the main chain direction was set to 35°. Then, the structural parameters of the FGE fragment itself were calculated, setting its average atomic density to 18 atoms / 100 ų, the main chain connection angles to be 112° and 113°, and the spatial angle to be 42°. The coordinate difference, angle difference, and angle offset were calculated item by item. The density difference was calculated. angular difference Angle offset To obtain the total offset value, weights are assigned to the three types of parameters. The results were derived from large-scale mutation experiment data, using multiple regression analysis to determine the contribution of each parameter to structural stability. Then the total offset value Perform this calculation on all candidate fragments to generate a set of three-parameter geometric offsets.

[0082] S402: Based on the numbering sequence of each path group in the three-parameter geometric offset set, extract the local conformation and stability values ​​of the corresponding residues. Within a fixed range, combine the fitting accuracy value and stability value to form a co-change sequence. Determine whether its change trend is within the fluctuation rate change threshold range, obtain the numbered path that meets the trend co-change condition, and obtain the co-conformation trend sequence.

[0083] Based on the three-parameter geometric offset set, the number sequence of FGE path groups is extracted, and their local conformation and stability values ​​are retrieved. Conformation is determined by Laplace plot regions; all FGEs are located within the allowable region. Stability is evaluated using statistical energy scores based on distance potential; lower scores indicate greater stability. Let the stability score of FGE be -25.8. Another candidate segment, WYE, has an offset value of 3.520 and a stability score of -28.4. The fitting accuracy is defined as the reciprocal of the total offset value, i.e., ... Then the fitting accuracy of FGE is The fitting accuracy of WYE is The fitting accuracy and stability values ​​of a series of candidate fragments are combined to form a co-variation sequence. The trend of this variation is then assessed to determine if it falls within a threshold range for the rate of change in fluctuation. This threshold is determined by analyzing the relationship between sequence variation and stability changes in the natural protein family and is set as the slope of a linear regression. When the fitting accuracy increases (the offset decreases), the stability score should decrease accordingly (becoming more stable). A linear fit is performed between the points (0.243, -25.8) and (0.284, -28.4), and the slope is... Since a lower stability score is better, a negative slope represents a positive correlation, satisfying the trend coordination condition. The numbered paths FGE and WYE that meet the condition are obtained, resulting in the coordinated conformation trend sequence.

[0084] S403: Call the path number in the co-conformation trend sequence, retrieve the corresponding offset combination value in the three-parameter geometric offset set, filter the sequence paths whose total offset value is less than the geometric compatibility threshold, and integrate the filtered paths by number to establish a list of adaptive sequence fragments.

[0085] The path numbers FGE and WYE in the obtained co-conformation trend sequence are retrieved, and their corresponding offset combination values ​​in the three-parameter geometric offset set are found to be 4.115 and 3.520, respectively. A geometric compatibility threshold is then set. The threshold was determined by performing a retrospective analysis of 500 validated successful protein design cases, calculating the geometric offset values ​​of their replacement fragments, and selecting an upper limit of the offset value that could cover 95% of the successful cases. The experiment set this threshold as follows: The sequence paths with a total offset value less than the threshold are filtered. It can be seen from the comparison that the offset value of FGE is 4.115, which is greater than 4.0, and does not meet the condition. The offset value of WYE is 3.520, which is less than 4.0, and meets the condition. Therefore, the WYE path is numbered and integrated to establish an adaptive sequence fragment list.

[0086] Please see Figure 6 The steps to obtain S5 are as follows:

[0087] S501: Based on the numbering order of each highly matched sequence fragment in the adaptive sequence fragment list, splice them according to the node number corresponding to them in the original structural coordinates, call the side chain charge data and contact surface projection area value of the residues at the splicing interface, extract the electronic distribution and physical contact range information of the residue contact area, and obtain the fragment splicing interface parameter set;

[0088] Based on the list of adaptive sequence fragments, the sequence fragment WYE with the highest matching degree was selected and spliced ​​according to its corresponding node numbering order in the original structural coordinates (201, 202, 203), replacing the original LGE fragment. Next, the side chain charge data and the projected area value of the contact surface of the residues at the splicing interface were called. Interface 1 is the connection between the original residue 200 and the first residue W (tryptophan) of the WYE fragment, and interface 2 is the connection between the last residue E (glutamate) of the WYE fragment and the original residue 204. At interface 1, residue 200 is aspartic acid (D, negatively charged), residue 204 is lysine (K, positively charged), W is neutral, and E is negatively charged. The side chain charges of D and W at interface 1 are extracted as -1 and 0, respectively, and the side chain charges of E and K at interface 2 are -1 and +1, respectively. The projected area of ​​the contact surface is obtained by calculating the projection of the two residue side chain atoms on the contact plane. The area value of interface 1 is set to 45.2 Ų, and that of interface 2 is set to 58.6 Ų. This information together constitutes the fragment splicing interface parameter set, as shown in the table below.

[0089] Table 2: Fragment splicing interface parameter table

[0090] Interface number Linking residue pairs Side chain charge pairs Projected area of ​​the contact surface (Ų) 1 D200-W201' (-1,0) 45.2 2 E203'-K204 (-1,+1) 58.6

[0091] As shown in Table 2, this table lists the key physicochemical parameters of the two interface locations after splicing.

[0092] S502: Based on the contact surface information of each interface residue in the fragment splicing interface parameter set, analyze the matching of the electron repulsion region and the spatial coordination region in the residue docking combination, detect the main chain rotation angle value of each interface connection site, calculate the angle deviation value between the angle and the original main chain direction, and establish a connection angle offset value table.

[0093] Analyzing the contact surface information of each interface residue in the fragment splicing interface parameter set, at interface 2, the negative charge of residue E and the positive charge of residue K form a favorable salt bridge, belonging to spatial coordination region matching. At interface 1, the negative charge of D and the neutral bulky side chain of W do not show obvious charge repulsion, but the van der Waals radius needs to be checked to avoid steric hindrance conflicts. Through atomic coordinate calculation, no interatomic spacing less than 0.8 times the sum of their van der Waals radii was found, indicating no serious collisions. Subsequently, the main chain rotation angle values ​​at each interface connection site were detected, especially the peptide bond plane angle ω. Assuming the measured ω angle of D200-W201′ at interface 1 is 178.5°, and the ω angle of E203′-K204 at interface 2 is -177.0°, the angular deviation value of this angle from the ideal trans peptide bond (180°) or the original main chain direction was calculated. The deviation of interface 1 is... The deviation of interface 2 is Record these deviation values ​​along with the deviations of the angles φ and ψ, and establish a table of connection angle offset values.

[0094] Table 3: Values ​​of Connection Angle Offset

[0095] Interface number ω-angle deviation (°) φ angle deviation (°) ψ angle deviation (°) 1 1.5 5.2 8.1 2 3.0 6.8 4.5

[0096] As shown in Table 3, this table quantifies the perturbation of the main chain geometry caused by the splicing operation.

[0097] S503: Call the connection points whose offset values ​​in the connection angle offset value table meet the directional continuity range, extract the local conformation data of the key connection points and perform directional boundary difference correction processing, combine and integrate the directional data of the corrected connection nodes with the original sequence fragments to generate the target protein structure design sequence.

[0098] The offset values ​​recorded in the connection angle offset value table are called and compared with the preset directional continuity range. This range is determined by the statistical distribution of main chain angles in the high-resolution crystal structure database, and is set as follows: ω angle deviation is less than 5° and φ / ψ angle deviation is less than 10°. As shown in Table 3, all offset values ​​of interface 1 and interface 2 are within their respective threshold ranges. Therefore, it is determined that both connection points meet the directional continuity requirements and no boundary correction is required. If we assume that the φ angle deviation of interface 1 is 12.7°, which exceeds the 10° threshold, then directional boundary difference correction processing needs to be performed on the connection point. The specific execution process of this correction processing is as follows: First, using the ψ angle of residue D200 and the φ angle of residue W201′ at interface 1 as adjustable variables, a combined fine-tuning is performed within a range of ±5° with a step size of 1° to generate a series of candidate conformations. For each candidate conformation... For example, the deviation of the φ angle of W201′ from the original main chain direction is recalculated, and conformations with deviations still greater than 10° are discarded. For conformations that meet the angular deviation requirements, their local comprehensive energy fraction is further calculated. This fraction consists of two weighted components. The weights are derived from regression analysis of the contribution of each energy term to structural stability in large-scale molecular dynamics simulations. The van der Waals term is set to a weight of 0.6, and the electrostatic term to 0.4. The first is the van der Waals collision penalty term, which is calculated by the distance between the side chain atoms of D200 and W201′. When the distance is less than 0.9 times the sum of the van der Waals radii, an energy penalty value inversely proportional to the square of the distance difference is assigned. For example, the initial collision penalty value is 15.3. The second is the electrostatic interaction energy, which is calculated based on the -1 charge of D200 and the local charge distribution of the side chain atoms of W201′. The initial value is -5.8, and the initial comprehensive energy fraction is... After traversing all candidate conformations, the conformation that minimizes the local integrated energy fraction and satisfies the angle deviation constraint is selected as the correction result. For example, when the ψ angle of D200 is adjusted by +2° and the φ angle of W201′ is adjusted by -3°, the φ angle deviation of W201′ decreases to 9.5°, the van der Waals collision penalty decreases to 2.1 due to the increase in interatomic spacing, and the electrostatic interaction energy becomes -6.2 due to relative position optimization. At this time, the new integrated energy fraction is... This conformation was adopted because it had the lowest overall energy fraction. The final angle data of the corrected linker (D200-W201′) was updated in the structural coordinates and combined with the WYE sequence fragment and the rest of the protein to generate the target protein structure design sequence.

[0099] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A multi-modal deep learning intelligent design method based on protein sequence and structure, characterized in that, The method comprises the following steps: S1: obtaining a 3Di structure string of a target protein, collecting three-dimensional coordinates, rotation angles and direction vectors of main chain nodes, extracting index numbers, hydrophobicity values and charge polarity of each residue in the target protein sequence, constructing a pairing relationship based on gradient amplitude and angle change characteristics, screening point position combinations, and generating a multi-modal collaborative feature node group; the multi-modal collaborative feature node group comprises collaborative feature points, associated domain gradient values and sequence domain angle change rates; S2: based on the multi-modal collaborative feature node group, extracting the main chain torsion angle change value, the residue potential difference value and the spatial density change rate corresponding to the node, analyzing the relative variation interval between the three types of values, screening the point set with a continuous gradient interval, calculating the spatial mean and standard deviation, and generating an embedding constraint condition set; the embedding constraint condition set is specifically a reference window coordinate mean value, a target main chain direction vector and a torsion angle change threshold value; S3: according to the window three-dimensional coordinate mean value and the main chain direction vector in the embedding constraint condition set, calling a candidate residue group of the same length, comparing the hydrophobicity values and the spatial direction vector angle of each residue, and generating a candidate residue sequence fragment group; the candidate residue sequence fragment group specifically refers to a hydrophobicity matched amino acid combination, a consistent number after verification, and a potential connection path; S4: based on the candidate residue sequence fragment group and the density distribution, the main chain connection angle and the spatial angle of the target embedding position, calculating the geometric offset of the three types of parameters, screening the sequence fragments with compatible structure, and generating an adaptive sequence fragment list; the adaptive sequence fragment list comprises a high-score compatibility sequence, geometric offset evaluation data and conformation stability fitting value.

2. The multi-modal deep learning intelligent design method based on protein sequence and structure of claim 1, wherein, The acquisition step of the multi-modal collaborative feature node group is: S101: obtaining a 3Di structure string of a target protein, collecting three-dimensional coordinate data, rotation angle values and direction vector coordinate information of main chain nodes in the structure, extracting index numbers of corresponding residues in the target protein sequence according to the spatial distribution order of each main chain node in the structure expression domain, calling hydrophobicity values and charge polarity values of corresponding residues in the database based on the index numbers, and generating a residue conformation attribute group; S102: based on the index numbers of residues in the residue conformation attribute group, calling gradient values and angle change rate data corresponding to the index numbers in the sequence and structure expression domains respectively, calculating the difference interval of gradient amplitude values and angle change rate values under each number, establishing a gradient-angle joint trajectory sequence according to the index number order of the residues, and obtaining a joint trajectory variation interval value; S103: according to the joint trajectory variation interval value, judging the adjacent points of the synchronous variation trend of the gradient amplitude value and the angle change rate value, screening out the point combination with synchronous growth or synchronous decay of the gradient, and jointly forming multi-modal node attributes of three-dimensional coordinates, hydrophobicity values and charge polarity values in the corresponding point combination, to generate a multi-modal collaborative feature node group. 3.The protein sequence and structure based multi-modal deep learning intelligent design method of claim 1, wherein, The acquisition step of the embedding constraint condition set is: S201: Based on the multi-modal collaborative feature node group, the torsion angle change value data of the node corresponding to the main chain residue is extracted, the potential value of the residue corresponding to the node is matched, the difference between the potential values of adjacent residues is compared with the sequence index, the spatial density change rate of the corresponding node is collected in combination with the three-dimensional coordinate range, the joint variation sequence is established according to the node number of the three values, and the joint structure variation parameter set is generated; S202: According to the distribution change interval of the three types of values in the joint structure variation parameter set in the node index sequence, the start and end positions of the continuous gradient change region are judged, the node index combination in which each parameter is simultaneously in the continuous unidirectional variation interval is screened, the node number set in the corresponding combination is obtained, and the continuous variation node set is established; S203: The three-dimensional coordinate set and direction vector value of the node contained in the continuous variation node set are called, the spatial coordinate mean value and direction vector standard deviation are calculated respectively, the node number group corresponding to the minimum standard deviation is selected as the structure embedding reference condition in all standard deviation results, and the embedding constraint condition set is established in combination with the mean value and vector data.

4. The multi-modal deep learning intelligent design method based on protein sequence and structure of claim 1, wherein, The obtaining step of the residue sequence fragment group is: S301: According to the window three-dimensional coordinate mean value and main chain direction vector value in the embedding constraint condition set, a candidate residue group of the same number is called, the hydrophobicity value of each residue in the candidate group is extracted respectively, the included angle value between the direction vector of each residue and the window direction vector is calculated, the corresponding matching matrix of the hydrophobicity difference value and the included angle value is obtained, and the hydrophobicity included angle matching data set is generated; S302: Based on the hydrophobicity included angle matching data set, the path combination in which the included angle value and the hydrophobicity difference value are simultaneously located in the included angle threshold interval and the hydrophobicity threshold interval is screened, the continuity of the residue number sequence in the screened path is judged, and the sequence consistency is verified by judging the index difference value of the residue number sequence to obtain the continuous sequence consistency path group; S303: The direction vector sequence and residue hydrophobicity gradient of all path combinations in the continuous sequence consistency path group are called, the trend direction of the direction change rate and the hydrophobicity change interval is extracted, the path combination in which the direction change trend and the hydrophobicity gradient direction are consistent is integrated, and the candidate residue sequence fragment group is established.

5. The multi-modal deep learning intelligent design method based on protein sequence and structure of claim 1, wherein, The obtaining step of the adaptive sequence fragment list is: S401: Based on the number information of each sequence fragment in the candidate residue sequence fragment group, the three-dimensional coordinate density value, the main chain connection angle value and the spatial included angle value corresponding to the target embedding position are called, the coordinate difference, the angle difference and the included angle offset are calculated item by item respectively with the structure parameters corresponding to each fragment, the total offset value of each fragment under the three types of parameters is obtained, and the three-parameter geometric offset set is generated; S402: According to the number sequence of each path group in the three-parameter geometric offset set, the local conformation and stability value of the corresponding residue are extracted, the fitting accuracy value and the stability value are combined into a collaborative change sequence in a fixed range, whether the change trend is within the fluctuation rate change threshold is judged, the number path meeting the trend collaborative condition is obtained, and the collaborative conformation trend sequence is obtained. S403: Call the path number in the conformational trend sequence, retrieve its corresponding offset combination value in the three-parameter geometric offset set, filter the sequence path with an offset value total less than the geometric compatibility threshold, and number and integrate the filtered path to establish an adaptive sequence fragment list.

6. The multi-modal deep learning intelligent design method based on protein sequence and structure of claim 1, wherein, The method further comprises: S5: According to the high matching sequence fragments in the adaptive sequence fragment list, splice the residue side chain charge and the contact surface projection area at the interface, identify the rotation angle deviation of the connection part, correct the direction continuity range and boundary by limiting, and integrate the adjusted residue combination into the target protein structure design sequence; The target protein structure design sequence is specifically a complete spliced sequence, an optimized corrected interface conformation, and a final residue three-dimensional coordinate set.

7. The multi-modal deep learning intelligent design method based on protein sequence and structure of claim 6, wherein, The target protein structure design sequence is obtained by: S501: According to the numbering order of each high matching sequence fragment in the adaptive sequence fragment list, splice according to the corresponding node number in the original structure coordinates, call the side chain charge data and contact surface projection area value of the residue at the spliced interface, extract the electronic distribution and physical contact range information of the residue contact area, and obtain the fragment splicing interface parameter set; S502: Based on the contact surface information of each interface residue in the fragment splicing interface parameter set, analyze the matching of the electronic repulsion region and the space coordination region in the residue docking combination, detect the main chain rotation angle value of each interface connection part, calculate the angle deviation value of the angle and the original main chain direction, and establish a connection angle offset value table; S503: Call the connection points with offset values meeting the direction continuity range in the connection angle offset value table, extract the local conformation data of the key connection points and perform direction boundary difference correction processing, integrate the direction data of the corrected connection nodes with the original sequence fragment combination, and generate the target protein structure design sequence.

Images