A local-global template search method for multidomain proteins

Abstract

A local-global template search method of multi-domain protein, first, input the three-dimensional structure of each single domain of the multi-domain protein to be assembled, set the number R of templates to be screened in local evaluation; then, score the alignment of each template in the multi-domain protein structure library with each two consecutive domains, respectively calculate the local score of each two consecutive domains from N terminal to C terminal and from C terminal to N terminal, select the template with the highest local score as the optimal local template; then, again for each template in the multi-domain protein structure library, perform local evaluation by structure alignment of each domain and template through TM-align, and calculate the score of the template according to the local score of each single domain, select the top R templates with high score for global evaluation; in global evaluation, the alignment of different domains is not allowed to overlap, select the template with the highest global score as the optimal global template; finally, output the optimal local template and the final full-length template. The application improves the search precision of the template.

Description

Technical Field

[0001] This invention relates to the fields of bioinformatics, intelligent optimization, and computer applications, and in particular to a local-global template search method for multi-domain proteins. Background Technology

[0002] Proteins are the material basis of all living systems, supporting almost all functions of life. The way proteins work and function largely depends on their unique three-dimensional structure. Most proteins in nature contain multiple folded units (or domains). Although each domain can perform biological functions independently, multiple domains can combine to perform more complex biological functions. Therefore, many biological functions depend on the interactions between domains.

[0003] Currently, only 34.7% of the 608,044 protein chains in the Protein Data Bank (PDB) contain multiple domains. Similarly, almost all advanced protein structure prediction methods, such as I-TASSER, Quark, and Rosetta, are optimized for single-domain protein modeling in terms of both force field design and conformational space search. These methods have led to a significant gap between the ability to determine single-domain structures and the high demand in the biomedical community for high-resolution multi-domain protein structures. While existing single-domain protein structure prediction methods (such as QUARK, I-TASSER, and ROSETTA) have achieved a certain level of accuracy in predicting the structure of single-domain proteins, multi-domain protein structure modeling remains a significant problem neglected by mainstream computational biology, compared to the extensive research and rapid progress in single-domain tertiary folding.

[0004] Many single-domain protein structure prediction methods cannot be directly applied to multi-domain protein structure prediction. Currently, commonly used multi-domain protein prediction methods fall into two categories: those that predict the structure of a single-domain protein using single-domain prediction methods, and then sample the hinge regions between domains or perform rigid docking of the domains. During domain docking, template guidance can improve prediction accuracy. Most template search methods are based on threading methods, starting from the amino acid sequence and using information such as secondary structure and solvent accessibility to search. However, the assembly process of multi-domain proteins relies solely on amino acid sequence information without searching for templates based on the three-dimensional structure information of individual domains. This results in the inability to obtain accurate local and global orientation information between domains, leading to lower assembly accuracy for multi-domain proteins.

[0005] Therefore, existing multi-domain protein template search methods have shortcomings in obtaining local and global orientation information of structural domains and need to be improved. Summary of the Invention

[0006] To overcome the shortcomings of existing multi-domain protein template search methods, such as the inability to obtain accurate local and global orientation information between domains and the low quality of templates, this invention provides a local-global template search method for multi-domain proteins that can obtain local and global orientation information between domains and has high template quality.

[0007] The technical solution adopted by this invention to solve its technical problem is:

[0008] A local-global template search method for multi-domain proteins, the method comprising the following steps:

[0009] 1) Input the three-dimensional structures of all single domains of the multi-domain protein to be assembled;

[0010] 2) Setting parameters: The number of templates to be screened after local evaluation, R;

[0011] 3) Scoring is performed on the alignment of each template with every two consecutive domains in the multi-domain protein structure library, as follows:

[0012] 3.1) Using the protein structure alignment tool TM-align (https: / / zhanggroup.org / TM-align / ), the first single domain was structurally aligned with the template protein starting from the N-terminus of the template, and the alignment score (TM-score) between the single domain and the template was calculated. NC，1 ;

[0013] 3.2) Ignore all residues in the template that align with the first single domain, and calculate the alignment score (TM-score) between the template and the second single domain according to step 3.1). NC，2 ;

[0014] 3.3) Calculate the local score LScore1 when aligning the template from N to C:

[0015]

[0016] 3.4) Using the protein structure alignment tool TM-align, starting from the C-terminus of the template, the first single domain is structurally aligned with the template, and the alignment score (TM-score) between the single domain and the template is calculated. CN，1 ;

[0017] 3.5) Ignore all residues in the template that align the first single domain, and calculate the alignment score (TM-score) between the deleted template and the second single domain. CN，2 ;

[0018] 3.6) Calculate the local score LScore2 when aligning the template from C to N:

[0019]

[0020] 3.7) Select the higher of LScore1 and LScore2 as the final local score LScore of the template with respect to the two continuous domains;

[0021] 4) For every two consecutive domains, sort all templates from high to low according to their corresponding LScores, and select the template with the highest score as its optimal local template.

[0022] 5) Perform local evaluation on each template protein in the multi-domain protein structure library, as follows:

[0023] 5.1) Align each single domain with the template using the protein structure alignment tool TM-align;

[0024] 5.2) Calculate the alignment score between each single domain and the template, and define the harmonic mean of the alignment scores of all domains as the template score (TM-score). h :

[0025]

[0026] Where N dom It is the total number of single domains to be assembled, TM-score d It is the comparison score between the d-th single domain and the template structure;

[0027] 6) Calculate the TM-score for each template based on the score calculated in step 5.2). h The templates with the highest R values ​​are selected and ranked from highest to lowest.

[0028] 7) Perform a global evaluation of the top-ranked R templates, as follows:

[0029] 7.1) Following step 5.2), use TM-align to align each field with the template from the N end to the C end. During this alignment process, overlap between different fields is not allowed. The resulting template score (TM-score) is then obtained. h1 ;

[0030] 7.2) Following step 5.2), use TM-align to align each field with the template from C to N. During this alignment process, overlap between different fields is not allowed. The resulting template score (TM-score) is then obtained. h2 ;

[0031] 7.3) Rank the scores from highest to lowest in steps 7.1) and 7.2), and select the template with the highest score as the optimal global template, as follows:

[0032] 7.3.1) If the optimal global template cannot cover all domains or the alignment score of at least one single domain is less than the preset threshold, then the structural domain closest to the N end will be used as the split point, and the structural domains at the N end and C end will be independently searched for templates respectively. Then, the structural domains at the split point will be aligned, and the templates with the highest scores from the independent search will be connected to generate the final full-length template.

[0033] 7.3.2) If the optimal global template can cover all domains, that is, the alignment score is greater than or equal to the preset threshold, then the template is the final full-length template;

[0034] 8) Output the optimal local template and the final full-length template.

[0035] Furthermore, in 7.3), the preset threshold is 0.5.

[0036] The technical concept of this invention is as follows: First, input the three-dimensional structure of each domain of the multi-domain protein to be assembled, and set the number of templates R to be screened in the local evaluation; then, score the alignment of each template with each pair of consecutive domains in the multi-domain protein structure library, calculate the local score for each pair of consecutive domains from the N-terminus to the C-terminus and from the C-terminus to the N-terminus, and select the template with the highest local score as the optimal local template; then, for each template in the multi-domain protein structure library, perform local evaluation by aligning each domain with the template using TM-align, and calculate the template score based on the local score of each domain, and screen the top R templates for global evaluation; wherein, the alignment of different domains in the global evaluation is not allowed to overlap, and the template with the highest global score is selected as the optimal global template; finally, output the optimal local template and the final full-length template.

[0037] The beneficial effects of this invention are as follows: on the one hand, by evaluating the template from local to global, and searching for the template starting from the three-dimensional structure of a single-domain protein, local and global directional information between domains can be obtained; on the other hand, by evaluating local and global templates, the search accuracy of the template can be improved. Attached Figure Description

[0038] Figure 1 This is a basic flowchart of local template search.

[0039] Figure 2 This is a basic flowchart of global template search.

[0040] Figure 3 This is an assembly diagram of the optimal global template to the final full-length template.

[0041] Figure 4 It is the three-dimensional structure of the final full-length template. Detailed Implementation

[0042] The present invention will now be further described with reference to the accompanying drawings.

[0043] Reference Figures 1-4 A local-global template search method for multi-domain proteins includes the following steps:

[0044] 1) Input the three-dimensional structures of all single domains of the multi-domain protein to be assembled;

[0045] 2) Setting parameters: The number of templates to be screened after local evaluation, R;

[0046] 3) Scoring is performed on the alignment of each template with every two consecutive domains in the multi-domain protein structure library, as follows:

[0047] 3.1) Using the protein structure alignment tool TM-align (https: / / zhanggroup.org / TM-align / ), the first single domain was structurally aligned with the template protein starting from the N-terminus of the template, and the alignment score (TM-score) between the single domain and the template was calculated. NC，1 ;

[0048] 3.2) Ignore all residues in the template that align with the first single domain, and calculate the alignment score (TM-score) between the template and the second single domain according to step 3.1). NC，2 ;

[0049] 3.3) Calculate the local score LScore1 when aligning the template from N to C:

[0050]

[0051] 3.4) Using the protein structure alignment tool TM-align, starting from the C-terminus of the template, the first single domain is structurally aligned with the template, and the alignment score (TM-score) between the single domain and the template is calculated. CN，1 ;

[0052] 3.5) Ignore all residues in the template that align the first single domain, and calculate the alignment score (TM-score) between the deleted template and the second single domain. CN，2 ;

[0053] 3.6) Calculate the local score LScore2 when aligning the template from C to N:

[0054]

[0055] 3.7) Select the higher of LScore1 and LScore2 as the final local score LScore of the template with respect to the two continuous domains;

[0056] 4) For every two consecutive domains, sort all templates from high to low according to their corresponding LScores, and select the template with the highest score as its optimal local template.

[0057] 5) Perform local evaluation on each template protein in the multi-domain protein structure library, as follows:

[0058] 5.1) Align each single domain with the template using the protein structure alignment tool TM-align;

[0059] 5.2) Calculate the alignment score between each single domain and the template, and define the harmonic mean of the alignment scores of all domains as the template score (TM-score). h :

[0060]

[0061] Where N dom It is the total number of single domains to be assembled, TM-score d It is the comparison score between the d-th single domain and the template structure;

[0062] 6) Calculate the TM-score for each template based on the score calculated in step 5.2). h The templates with the highest R values ​​are selected and ranked from highest to lowest.

[0063] 7) Perform a global evaluation of the top-ranked R templates, as follows:

[0064] 7.1) Following step 5.2), use TM-align to align each field with the template from the N end to the C end. During this alignment process, overlap between different fields is not allowed. The resulting template score (TM-score) is then obtained. h1 ;

[0065] 7.2) Following step 5.2), use TM-align to align each field with the template from C to N. During this alignment process, overlap between different fields is not allowed. The resulting template score (TM-score) is then obtained. h2 ;

[0066] 7.3) Rank the scores from highest to lowest in steps 7.1) and 7.2), and select the template with the highest score as the optimal global template, as follows:

[0067] 7.3.1) If the optimal global template cannot cover all domains or the alignment score of at least one single domain is less than 0.5, then the structural domain closest to the N end will be used as the split point, and independent templates will be searched for the structural domains at the N end and the C end respectively. Then, the structural domains at the split point will be aligned, and the templates with the highest independent search scores will be connected to generate the final full-length template.

[0068] 7.3.2) If the optimal global template can cover all domains, that is, the alignment score is greater than or equal to 0.5, then the template is the final full-length template;

[0069] 8) Output the optimal local template and the final full-length template.

[0070] This embodiment uses the 230-byte sequence of the multi-domain protein 1fx7A, which contains three domains, as an example. A local-global template search method for multi-domain proteins includes the following steps:

[0071] 1) Input the three-dimensional structures of all single domains of the multi-domain protein to be assembled;

[0072] 2) Set parameters: Number of templates to be screened after local evaluation R = 500;

[0073] 3) Scoring is performed on the alignment of each template with every two consecutive domains in the multi-domain protein structure library, as follows:

[0074] 3.1) Using the protein structure alignment tool TM-align (https: / / zhanggroup.org / TM-align / ), the first single domain was structurally aligned with the template protein starting from the N-terminus of the template, and the alignment score (TM-score) between the single domain and the template was calculated. NC，1 ;

[0075] 3.2) Ignore all residues in the template that align with the first single domain, and calculate the alignment score (TM-score) between the template and the second single domain according to step 3.1). NC，2 ;

[0076] 3.3) Calculate the local score LScore1 when aligning the template from N to C:

[0077]

[0078] 3.4) Using the protein structure alignment tool TM-align, starting from the C-terminus of the template, the first single domain is structurally aligned with the template, and the alignment score (TM-score) between the single domain and the template is calculated. CN，1 ;

[0079] 3.5) Ignore all residues in the template that align the first single domain, and calculate the alignment score (TM-score) between the deleted template and the second single domain. CN，2 ;

[0080] 3.6) Calculate the local score LScore2 when aligning the template from C to N:

[0081]

[0082] 3.7) Select the higher of LScore1 and LScore2 as the final local score LScore of the template with respect to the two continuous domains;

[0083] 4) For every two consecutive domains, sort all templates from high to low according to their corresponding LScores, and select the template with the highest score as its optimal local template.

[0084] 5) Perform local evaluation on each template protein in the multi-domain protein structure library, as follows:

[0085] 5.1) Align each single domain with the template using the protein structure alignment tool TM-align;

[0086] 5.2) Calculate the alignment score between each single domain and the template, and define the harmonic mean of the alignment scores of all domains as the template score (TM-score). h :

[0087]

[0088] Where N dom It is the total number of single domains to be assembled, TM-score d It is the comparison score between the d-th single domain and the template structure;

[0089] 6) Calculate the TM-score for each template based on the score calculated in step 5.2). h The templates with the highest R values ​​are selected and ranked from highest to lowest.

[0090] 7) Perform a global evaluation of the top-ranked R templates, as follows:

[0091] 7.1) Following step 5.2), use TM-align to align each field with the template from the N end to the C end. During this alignment process, overlap between different fields is not allowed. The resulting template score (TM-score) is then obtained. h1 ;

[0092] 7.2) Following step 5.2), use TM-align to align each field with the template from C to N. During this alignment process, overlap between different fields is not allowed. The resulting template score (TM-score) is then obtained. h2 ;

[0093] 7.3) Rank the scores from highest to lowest in steps 7.1) and 7.2), and select the template with the highest score as the optimal global template, as follows:

[0094] 7.3.1) If the optimal global template cannot cover all domains or the alignment score of at least one single domain is less than 0.5, then the structural domain closest to the N end will be used as the split point, and independent templates will be searched for the structural domains at the N end and the C end respectively. Then, the structural domains at the split point will be aligned, and the templates with the highest independent search scores will be connected to generate the final full-length template.

[0095] 7.3.2) If the optimal global template can cover all domains, that is, the alignment score is greater than or equal to 0.5, then the template is the final full-length template;

[0096] 8) Output the optimal local template and the final full-length template.

[0097] Using the 230-byte multidomain protein 1fx7A containing three domains as an example, the optimal global template for this multidomain protein was obtained by assembling it using the above method. The global alignment score between 1fx7A and 4o6jA was 0.76. The three-dimensional structure diagram of the optimal global template 4o6jA is shown below. Figure 4 As shown, the optimal local template for domain 1 and domain 2 is 2x4hA, and the optimal local template for domain 2 and domain 3 is 5cviB.

[0098] The above description is based on the template search effect obtained by the present invention using 1fx7A protein as an example. It does not limit the scope of the present invention. Various modifications and improvements made to it without departing from the scope of the basic content of the present invention should not be excluded from the protection scope of the present invention.

Claims

1. A local-global template search method for multidomain proteins, characterized in that, The method comprises the following steps: 1) input the three-dimensional structure of all single domains of the multi-domain protein to be assembled; 2) set parameters: the number of templates to be screened after local evaluation R; 3) score the alignment of each template in the multi-domain protein structure library with each two consecutive domains, the process is as follows: 3.1) By protein structure alignment tool TM-align, structure alignment of the first single domain with template protein from N-terminal of template, calculate the alignment score TM-score between the single domain and template NC，1 ; 3.2) Ignore all residues in the template that align to the first single domain, calculate the alignment score TM-score between the template and the second single domain according to step 3.1) NC，2 ; 3.3) calculate the local score LScore1 when the template is aligned from N to C terminal: 3.4) Calculate the alignment score TM-score between the first single domain and the template by protein structure alignment tool TM-align, starting from the C-terminus of the template CN，1 ; 3.5) Calculate the alignment score TM-score between the truncated template and the second single domain, ignoring all residues in the template that align to the first single domain CN，2 ; 3.6) calculate the local score LScore2 when the template is aligned from C to N terminal: 3.7) select the higher value of LScore1 and LScore2 as the final local score LScore of the template for the two consecutive domains; 4) for each two consecutive domains, sort all templates from high to low according to the corresponding LScore, and select the template with the highest score as the optimal local template; 5) perform local evaluation on each template protein in the multi-domain protein structure library, the process is as follows: 5.1) align each single domain with the template by protein structure alignment tool TM-align; 5.2) Calculate the alignment score between each single domain and the template, define the harmonic mean of all domain alignment scores as the score of this template, TM-score h : where N dom is the total number of single domains to be assembled, TM-score d is the alignment score between the dth single domain and the template structure; 6) Select the top R templates from the list of templates ranked by TM-score h , from high to low. 7) perform global evaluation on the top R templates, the process is as follows: 7.1) According to step 5.2), align each domain with the template from N- to C- terminus using TM-align, in which the alignment between different domains is not allowed to overlap, resulting in the score of the template in this case, TM-score h1 ; 7.2) According to step 5.2), align each domain with the template from C-terminus to N- terminus using TM-align, in which the alignment between different domains is not allowed to overlap, resulting in the score of the template in this case, TM-score h2 ; 7.3) rank the scores in steps 7.1) and 7.2) from high to low, and select the template with the highest score as the optimal global template, as follows: 7.3.1) if the optimal global template cannot cover all domains or the alignment score of at least one single domain is less than a preset threshold, then take the domain closest to the N terminal as the segmentation point, search for templates independently for the N terminal and C terminal domains respectively, and then connect the templates with the highest scores searched independently according to the structure domain of the segmentation point to generate the final full-length template; 7.3.2) if the optimal global template can cover all domains, i.e. the alignment score is greater than or equal to the preset threshold, then the template is the final full-length template; 8) output the optimal local template and the final full-length template.

2. A method for local-global template search of multi-domain proteins as claimed in claim 1 wherein, In the 7.3), the preset threshold is 0.5.

Images