Protein complex structure similar template searching method

Abstract

A protein complex structure similar template searching method comprises the following steps: firstly, constructing a protein complex template database and a fusion template database; inputting a single-chain structure of the target protein, searching a global template and a local template in the compound template database according to the structure, and searching the global template and the local template again in the fusion template database if the quality of the searched templates is poor; and finally, selecting an optimal template obtained by searching in the two databases as a structural similar template of the target protein. The invention provides the protein complex structure similar template searching method with relatively high precision.

Description

Technical Field

[0001] The invention belongs to the fields of bioinformatics, intelligent optimization and computer application, and in particular relates to a method for searching similar templates of protein complex structures. Background Art

[0002] For most target proteins, it is more reliable and more convenient to obtain them through sequence alignment or sequencing based on sequence homology or structural similarity. Especially when the target protein has significant homology with the template, the prediction results are generally highly credible. This makes template-based structure prediction methods more popular in practical applications. However, the number of known protein complex structures in PDB is far less than that of monomeric proteins, resulting in a relatively small number of template structures required for prediction or even missing. Marsh et al. published a paper in Cell and Pasek et al. published a paper in Bioinformatics, pointing out that proteins with multiple domains (i.e., multi-domains) are mainly derived from a series of gene fusion and fission events, that is, most multi-domain proteins are formed by single-chain fusion and fission of protein complexes in the genome, which can reliably infer the inter-chain information of protein complexes from multi-domain proteins. Therefore, the multi-domain protein template library can be used as a supplement to the complex template library to construct a structure-similar template library and improve the integrity of the database.

[0003] At present, there are many methods for searching templates on the market. These methods include homology modeling methods such as SWISS-MODEL developed by Schwede Laboratory, Modeller developed by Sali Laboratory, and I-TASSER prediction method based on threading developed by Zhang Laboratory. Although these methods have achieved many successes, there are still some problems (such as incomplete template databases and low accuracy of redundant and protein complex template searches) that limit the accuracy and efficiency of existing algorithms; secondly, most algorithms require manual processing of some intermediate processes, which affects the use of users in non-structural biology fields.

[0004] Therefore, existing methods for searching protein complex templates have defects in accuracy and automation and need to be improved. Summary of the invention

[0005] In order to overcome the shortcomings of the existing methods for searching protein complex templates, such as low precision and the need for manual intervention during the execution process, the present invention provides a method for searching protein complex structure similar templates with high precision.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] A method for searching for templates similar to protein complex structures, the method comprising the following steps:

[0008] 1) Input the single-chain structures of the protein complex to be searched;

[0009] 2) Setting parameters: minimum number of single-chain residues N min , sequence similarity coefficient threshold S thresholds , structural similarity coefficient threshold T thresholds , template matching percentage C percentage , template structure similarity threshold G thresholds ;

[0010] 3) constructing a complex template database;

[0011] 4) Building a fusion template database;

[0012] 5) Search for similar templates from the complex template database; if the best template found has a score G score Less than G thresholds , then execute step 6) otherwise this template is the final template;

[0013] 6) Search for similar templates from the fusion template database and compare them with the best template searched in step 5), and take the template with the highest score as the final template.

[0014] Further, the process of 3) is as follows:

[0015] 3.1) Select all protein complexes from the PDB database;

[0016] 3.2) Remove residues less than N in each complex min A single chain;

[0017] 3.3) removing single strands in each complex that have no contact with other strands;

[0018] 3.4) According to sequence similarity S thresholds CD-HIT was used to remove redundant proteins;

[0019] 3.5) Use USalign to remove structural similarity with TM-score greater than T thresholds protein complexes;

[0020] 3.6) Remove all single-chain structures in the database again.

[0021] Further, the process of 4) is as follows:

[0022] 4.1) Select all multi-domain proteins from the PDB database;

[0023] 4.2) According to sequence similarity S thresholds CD-HIT was used to remove redundant proteins;

[0024] 4.3) Use CD-HIT to sort the remaining proteins in CATH and SCOPe that have a sequence similarity less than S thresholds The multi-domain proteins were added to the database;

[0025] 4.4) Use TMalign to align sequences with similarity greater than S in PDB, CATH and SCOPe thresholds But TM-score is smaller than T thresholds The multi-domain proteins were added to the database.

[0026] Furthermore, the process of 5) is as follows:

[0027] 5.1) Searching for global templates: Use TM-align to align each chain in the complex template database. After each chain is aligned, remove the matched residues in the template and then match the next chain. Record the percentage of the number of matched residues in each chain to the number of residues in that chain. If the matching percentage is greater than C percentage , then the chain is considered to have been matched by the template, and the templates that all chains match are selected, and G is selected. score The highest template is used as the global template, where G score The calculation formula is as follows:

[0028]

[0029] N chain The number of all single chains in the input complex structure, TM-score d is the TM-score of the d-th chain;

[0030] 5.2) Searching for local templates: If the template cannot cover all chains, then according to the sliding window process, each two adjacent chains are searched for the global templates of these two chains in the complex template database in turn, and then the global templates corresponding to each group of different chain numbers are connected to the two templates based on the same chain number, so as to obtain the local template of the complex;

[0031] 5.3) Determine the best local template and the G corresponding to the global template score With G thresholds Compare, if greater than G thresholds , the template with the highest score is the final template of the complex, otherwise, proceed to step 6).

[0032] The technical concept of the present invention is as follows: first, a protein complex template database and a fusion template database are constructed; then the single-chain structure of the target protein is input, and the global template and the local template are searched for this structure in the complex template database. If the quality of the searched template is poor, the global template and the local template are searched again in the fusion template library; finally, the best template searched in the two databases is selected as the structurally similar template of the target protein.

[0033] The beneficial effects of the present invention are as follows: using multi-domain protein data as a supplement to construct a structure-similar template library to improve the integrity of the database; and using two methods of searching global templates and local templates to improve the accuracy of searching templates. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 is the basic flow chart of this method.

[0035] Figure 2 It is a flow chart of the construction of the complex template database.

[0036] Figure 3 It is a flow chart for building a fusion template database.

[0037] Figure 4 They are two single chains of the protein complex 3R8I.

[0038] Figure 5 It is the best complex template 2I4J found by this method. DETAILED DESCRIPTION

[0039] The present invention will be further described below in conjunction with the accompanying drawings.

[0040] Reference Figure 1 to Figure 5 , a protein complex structure similar template search method, comprising the following steps:

[0041] 1) Input the single-chain structures of the protein complex to be searched;

[0042] 2) Setting parameters: minimum number of single-chain residues N min , sequence similarity coefficient threshold S thresholds , structural similarity coefficient threshold T thresholds , template matching percentage C percentage , template structure similarity threshold G thresholds ;

[0043] 3) Constructing a complex template database, the process is as follows:

[0044] 3.1) Select all protein complexes from the PDB database;

[0045] 3.2) Remove residues less than N in each complexmin A single chain;

[0046] 3.3) removing single strands in each complex that have no contact with other strands;

[0047] 3.4) According to sequence similarity S thresholds CD-HIT (https: / / github.com / weizhongli / cdhit / releases) was used to remove redundant proteins;

[0048] 3.5) Use USalign (https: / / zhanggroup.org / US-align / ) to remove structural similarity TM-

[0049] score is greater than T thresholds protein complexes;

[0050] 3.6) Remove all single-chain structures from the database again;

[0051] 4) Build a fusion template database. The process is as follows:

[0052] 4.1) Select all multi-domain proteins from the PDB database;

[0053] 4.2) According to sequence similarity S thresholds CD-HIT (https: / / github.com / weizhongli / cdhit / releases) was used to remove redundant proteins;

[0054] 4.3) Use CD-HIT (https: / / github.com / weizhongli / cdhit / releases) to sort the remaining proteins in CATH and SCOPe that have a sequence similarity less than S thresholds The multi-domain proteins were added to the database;

[0055] 4.4) Use TMalign (https: / / zhanggroup.org / TM-align / ) to align the sequences in PDB, CATH and SCOPe with similarity greater than S thresholds But TM-score is smaller than T thresholds The multi-domain proteins were added to the database;

[0056] 5) Search for similar templates from the complex template database; if the best template found has a score G score Less than G thresholds , then execute step 6) otherwise this template is the final template; the process is as follows:

[0057] 5.1) Searching for global templates: Use TM-align (https: / / zhanggroup.org / TM-align / ) to align each chain in the complex template database. After each chain is aligned, remove the matched residues in the template and then match the next chain. Record the percentage of the number of matched residues in each chain to the number of residues in that chain. If the matching percentage is greater than C percentage , then the chain is considered to have been matched by the template, and the templates that all chains match are selected, and G is selected. score The highest template is used as the global template, where G score The calculation formula is as follows:

[0058]

[0059] N chain The number of all single chains in the input complex structure, TM-score d is the TM-score of the d-th chain;

[0060] 5.2) Searching for local templates: If the template cannot cover all chains, then according to the sliding window process, each two adjacent chains are searched for the global templates of these two chains in the complex template database in turn, and then the global templates corresponding to each group of different chain numbers are connected to the two templates based on the same chain number, so as to obtain the local template of the complex; taking the input as three chains (chain A, chain B, chain C) as an example, first search the database for the global template AB that can cover chain A and chain B and the global template BC that can cover chain B and chain C; then connect the two templates AB and BC according to the comparison of chain B to finally obtain the local template ABC of the three chains;

[0061] 5.3) Determine the best local template and the G corresponding to the global template score With G thresholds Compare, if greater than G thresholds , the template with the highest score is the final template of the complex, otherwise, go to step 6);

[0062] 6) Search for similar templates from the fusion template database and compare them with the best template searched in step 5), and take the template with the highest score as the final template.

[0063] This embodiment includes two single chains in the protein complex 3R8I as an embodiment, a method for searching for templates with similar structures of protein complexes, the method comprising the following steps:

[0064] 1) Input the single-chain structures of the protein complex to be searched;

[0065] 2) Setting parameters: minimum number of single-chain residues N min =30, sequence similarity coefficient threshold Sthresholds =70%, structural similarity coefficient threshold T thresholds =0.5, template matching percentage C percentage = 60%, template structure similarity threshold G thresholds＝ 0.65;

[0066] 3) Constructing a complex template database, the process is as follows:

[0067] 3.1) Select all protein complexes from the PDB database;

[0068] 3.2) Remove residues less than N in each complex min A single chain;

[0069] 3.3) removing single strands in each complex that have no contact with other strands;

[0070] 3.4) According to sequence similarity S thresholds CD-HIT (https: / / github.com / weizhongli / cdhit / releases) was used to remove redundant proteins;

[0071] 3.5) Use USalign (https: / / zhanggroup.org / US-align / ) to remove structural similarity TM-

[0072] score is greater than T thresholds protein complexes;

[0073] 3.6) Remove all single-chain structures from the database again;

[0074] 4) Build a fusion template database. The process is as follows:

[0075] 4.1) Select all multi-domain proteins from the PDB database;

[0076] 4.2) According to sequence similarity S thresholds CD-HIT (https: / / github.com / weizhongli / cdhit / releases) was used to remove redundant proteins;

[0077] 4.3) Use CD-HIT to sort the remaining proteins in CATH and SCOPe that have a sequence similarity less than S thresholds The multi-domain proteins were added to the database;

[0078] 4.4) Use TMalign (https: / / zhanggroup.org / TM-align / ) to align the sequences in PDB, CATH and SCOPe with similarity greater than S thresholdsBut TM-score is smaller than T thresholds The multi-domain proteins were added to the database;

[0079] 5) Search for similar templates from the complex template database; if the best template found has a score G score Less than G thresholds , then execute step 6) otherwise this template is the final template; the process is as follows:

[0080] 5.1) Search for global template: Use TM-align to align each chain in the complex template database

[0081] (https: / / zhanggroup.org / TM-align / ) for alignment. After each chain is aligned, the matched residues are removed from the template and then the next chain is matched. The percentage of the number of matched residues in each chain to the number of residues in that chain is recorded. If the matching percentage is greater than C percentage , then the chain is considered to have been matched by the template, and the templates that all chains match are selected, and G is selected. score The highest template is used as the global template, where G score The calculation formula is as follows:

[0082]

[0083] N chain The number of all single chains in the input complex structure, TM-score d is the TM-score of the d-th chain;

[0084] 5.2) Searching for local templates: If the template cannot cover all chains, then according to the sliding window process, each two adjacent chains are searched for the global templates of these two chains in the complex template database in turn, and then the global templates corresponding to each group of different chain numbers are connected to the two templates based on the same chain number, so as to obtain the local template of the complex;

[0085] 5.3) Determine the best local template and the G corresponding to the global template score With G thresholds Compare, if greater than G thresholds , the template with the highest score is the final template of the complex, otherwise, go to step 6);

[0086] 6) Search for similar templates from the fusion template database and compare them with the best template searched in step 5), and take the template with the highest score as the final template.

[0087] This example takes two single chains in the protein complex 3R8I as an example. The two single chain structures are as follows. Figure 4 As shown, the complex template structure 2I4J obtained by searching according to this method is as follows Figure 5 shown.

[0088] The above description is the search result obtained by the present invention using two single chains in the protein complex 3R8I as an example, and does not limit the scope of implementation of the present invention. Various modifications and improvements made thereto without departing from the scope involved in the basic content of the present invention should not be excluded from the protection scope of the present invention.

Claims

1. A method for searching for templates similar to protein complex structures, characterized in that: The method comprises the following steps: 1) Input the single-chain structures of the protein complex to be searched; 2) Setting parameters: minimum number of single-chain residues N min , sequence similarity coefficient threshold S thresholds , structural similarity coefficient threshold T thresholds , template matching percentage C percentage , template structure similarity threshold G thresholds ; 3) constructing a complex template database; 4) Building a fusion template database; 5) Search for similar templates from the complex template database; if the best template found has a score G score Less than G thresholds , then execute step 6) otherwise this template is the final template; 6) Search for similar templates from the fusion template database and compare them with the best template searched in step 5), and take the template with the highest score as the final template.

2. A method for searching for templates similar to protein complex structures according to claim 1, characterized in that: The process of 3) is as follows: 3.1) Select all protein complexes from the PDB database; 3.2) Remove residues less than N in each complex min A single chain; 3.3) removing single strands in each complex that have no contact with other strands; 3.4) According to sequence similarity S thresholds CD-HIT was used to remove redundant proteins; 3.5) Use USalign to remove structural similarity with TM-score greater than T thresholds protein complexes; 3.6) Remove all single-chain structures in the database again.

3. A method for searching for templates similar to protein complex structures according to claim 1 or 2, characterized in that: The process of 4) is as follows: 4.1) Select all multi-domain proteins from the PDB database; 4.2) According to sequence similarity S thresholds CD-HIT was used to remove redundant proteins; 4.3) Use CD-HIT to sort the remaining proteins in CATH and SCOPe that have a sequence similarity less than S thresholds The multi-domain proteins were added to the database; 4.4) Use TMalign to align sequences with similarity greater than S in PDB, CATH and SCOPe thresholds But TM-score is smaller than T thresholds The multi-domain proteins were added to the database.

4. A method for searching for templates similar to protein complex structures according to claim 1 or 2, characterized in that: The process of 5) is as follows: 5.1) Searching for global templates: Use TM-align to align each chain in the complex template database. After each chain is aligned, remove the matched residues in the template and then match the next chain. Record the percentage of the number of matched residues in each chain to the number of residues in that chain. If the matching percentage is greater than C percentage , then the chain is considered to have been matched by the template, and the templates that all chains match are selected, and G is selected. score The highest template is used as the global template, where G score The calculation formula is as follows: N chain The number of all single chains in the input complex structure, TM-score d is the TM-score of the d-th chain; 5.2) Searching for local templates: If the template cannot cover all chains, then according to the sliding window process, each two adjacent chains are searched for the global templates of these two chains in the complex template database in turn, and then the global templates corresponding to each group of different chain numbers are connected to the two templates based on the same chain number, so as to obtain the local template of the complex; 5.3) Determine the best local template and the G corresponding to the global template score With G thresholds Compare, if greater than G thresholds , the template with the highest score is the final template of the complex, otherwise, proceed to step 6).

Images