A method for targeted drug design based on chemical reactions and molecular building blocks

Abstract

The application discloses a kind of targeted drug design method based on chemical reaction and molecular building block, its characteristics are fully considered in the process of drug design synthesis problem, convert chemical building block into synthon, construct synthon database, then carry out virtual synthesis reaction in protein pocket to carry out drug design.In the molecular design stage, the method focuses on the structure of the target molecule, the binding ability with protein and the synthesizability to ensure that the designed molecule has good selectivity and biological activity.Synthon database provides molecular building block data for subsequent virtual synthesis algorithm, improves the synthesis feasibility of drug design molecules.Compared with the prior art, the application more fully considers the synthesizability of molecules in the drug design process and the restriction and constraint on fragments, accelerates the process of drug design, reduces the cost of drug development and improves the efficiency of drug development.

Description

Technical Field

[0001] This invention relates to the field of rational targeted drug design based on protein structure. Based on chemical reactions and molecular building blocks, it utilizes molecular growth technology to design targeted drugs based on protein structure. This method can consider the syntheticity of molecules while designing them, providing new ideas for the design and optimization of targeted drug molecules. Background Technology

[0002] Drug discovery is a complex and challenging task, requiring research and collaboration across multiple disciplines. Research into innovative drugs involves knowledge and technologies from chemistry, biology, pharmacology, informatics, and computer science, making it a long-term, costly, and high-risk endeavor. To become an effective drug, a molecule must not only possess the necessary pharmacological activity and selectivity but also appropriate absorption, distribution, metabolism, excretion, and safety properties. However, these properties are often subject to complex balances and interrelationships, making drug design and optimization even more challenging.

[0003] In drug design, in addition to considering the activity and properties of molecules, the ease of synthesizing new compounds is also an important factor. In practice, two methods are typically used to incorporate molecular synthesability: one approach is to assess synthesability after molecular design is complete using empirical rules (such as the number of chiral centers, the frequency of ring systems in the molecule's history, etc.) or by combining automated retrosynthetic analysis; the other approach is to directly select molecules that meet the project requirements from the chemical space composed of molecular building blocks and efficient organic chemical reactions, as such molecules naturally possess ease of synthesis.

[0004] Previously, molecular growth methods have been used in experiments to generate drugs with superior physicochemical properties. This has evolved from experimental methods to novel computational methods. However, existing molecular growth methods suffer from limitations such as limited data, difficulty in assessing syntheticity, operational complexity, and long processing times, leading to unnecessary time consumption and hindering the rapid discovery of clinically promising drugs. Summary of the Invention

[0005] The purpose of this invention is to provide a targeted drug design method based on chemical reactions and molecular building blocks. This method employs a constructed chemical reaction database, a synthesizer database, and a molecular growth algorithm for drug design. Through a molecular growth program, suitable growth points and fragments meeting specific conditions are identified, enabling virtual synthesis of molecules within a protein pocket, thus generating new drug-designed molecules. In the drug design process, this method allows medicinal chemists to quickly and easily explore suitable ligand design ideas using molecular growth when the target protein structure is available, while simultaneously considering the syntheticity of molecules. This accelerates the drug design process, reduces drug development costs, and improves drug development efficiency.

[0006] The objective of this invention is achieved as follows:

[0007] A targeted drug design method based on chemical reactions and molecular building blocks is characterized by the use of a constructed chemical reaction database, a synthon database, and a molecular growth algorithm for drug design. By converting chemical building blocks into synthons and performing virtual synthesis reactions in protein pockets, new drug design molecules are generated. A constructed small molecule fragment library and a molecular growth algorithm are then used for drug design. This design method identifies suitable growth sites through a molecular growth program to complete molecular growth and generate new drug design molecules.

[0008] The method includes the following specific steps:

[0009] Step 1: Search the literature for commonly used reactions in pharmaceutical chemical synthesis, extract chemical reaction templates, and construct a chemical reaction template database using SMIRKS;

[0010] Step 2: Based on the chemical reaction template database in Step 1, extract the SMARTS of the reactant portion and construct a reactant chemical environment database and a synthon preparation reaction database using SMARTS representation;

[0011] Step 3: Select molecules with 20 or fewer heavy atoms from the Enamine, BLDpharm, ChemDiv, WuxiAppTec, Chemieliva, StarshineChemical, Ambinter, and Mcule molecular building block databases; neutralize salt-containing molecules; and filter molecules containing elements other than C, H, O, N, S, P, Si, F, Cl, Br, I, and B.

[0012] Step 4: Based on the synthon preparation reaction database obtained in Step 2, the molecular building blocks obtained in Step 3 are converted into synthons using RDKit, and the reaction reagent information and reaction type information are recorded;

[0013] Step 5: Calculate the formal charge of the compound in Step 4 and the protonated state of the compound in Step 4 at pH=7;

[0014] Step 6: Use RDKit to generate the initial three-dimensional conformation for the compound from Step 5;

[0015] Step 7: Based on the reactant chemical environment database and synthon preparation reaction database obtained in Step 2, record the growth point atom ID, retained structure atom ID, deleted structure atom ID, molecular building blocks SMILES and reaction type of the three-dimensional molecules in Step 6, and construct the synthon database;

[0016] Step 8: Based on the reactant chemical environment database obtained in Step 2, calculate the reaction warhead on the seed compound or the smallest substructure obtained through the reaction in the database, i.e., the fragment growth of reagent replacement by one-step retrosynthetic analysis, and record the retained part atom ID and deleted part atom ID on the seed compound;

[0017] Step 9: Distance between the protein and the seed compound The inner part is defined as a protein pocket, and spatial relative positional relationship descriptors and fragment global property descriptors are calculated;

[0018] Step 10: Based on the spatial relative position relationship descriptor and fragment global property descriptor obtained in Step 9, search for a suitable synthesizer in the synthesizer database obtained in Step 7;

[0019] Step 11: Based on the synthon and seed compound obtained in Step 10, a series of new molecules that form specified interactions and meet the fragment requirements are obtained through virtual synthesis reaction at the growth point, and the results are visualized.

[0020] Furthermore, the construction of the chemical reaction template database using SMIRKS described in step 1 is constructed using SMARTS to represent the chemical environment constraints of reactants and products.

[0021] Furthermore, the construction of the reactant chemical environment database and synthon preparation reaction database using SMARTS in step 2 is based on the constraints on the chemical environment during the breaking of old bonds and the formation of new bonds in the reaction template represented by SMARTS, as well as the smallest fragment newly generated by the reaction.

[0022] Furthermore, the conversion of molecular building blocks into synthons in step 4 is based on the conversion of molecular building blocks into synthons by the reaction module in RDKit.

[0023] Furthermore, the growth point atom ID mentioned in step 7 is obtained by substructure matching based on RDKit, which yields the corresponding growth point atom ID, the ID of all atoms that are retained, and the ID of all atoms that are deleted.

[0024] Furthermore, step 8, which involves calculating the reaction warhead on the seed compound or the smallest substructure obtained through the reaction in the database (i.e., fragment growth via reagent replacement through one-step retrosynthetic analysis), is based on SMARTS in the RDKit matching database of chemical templates in step 1 and the chemical environment database of reactants in step 2. Specifically, this includes:

[0025] i) Extract the SMARTS of reactants from the reactant chemical environment database and use RDKit to find matching reaction warheads on seed compounds;

[0026] ii) Extract the smallest newly generated reaction unit from each reaction template in the chemical reaction template database, represent it in the form of SMARTS, find matching substructures on the seed compound using RDKit, and perform molecular growth by replacing one of the reagents in the reaction.

[0027] Furthermore, step 11 specifically includes: superimposing the seed compound and the synthon together according to the recorded growth point atom IDs, then removing some atoms and connecting the remaining atoms to complete molecular growth; based on the positions of the synthon and the seed compound at the growth points obtained in step 10, a series of new molecules that form specified interactions and meet the fragment requirements are obtained through virtual synthesis reactions, and the results are visualized.

[0028] Compared to other methods, the beneficial effects of this invention are as follows:

[0029] 1) By collecting commonly used and efficient chemical reactions in drug synthesis from the literature, using SMIRRKS to represent reaction templates and RDKit to carry out chemical reactions, molecular building blocks are transformed into synthons, and a database of structurally rich synthons is constructed for virtual synthetic reactions in protein pockets. This allows for consideration of syntheticity issues in the drug design process, accelerating the drug design process, reducing drug development costs, and improving drug development efficiency.

[0030] 2) The synthesizer database makes it easier to conduct virtual reactions in the protein pocket, allows for faster data searches, and results in better performance and shorter processing time.

[0031] 3) Compared with simple molecular growth, the combination of molecular fragment growth and molecular fragment replacement is achieved through virtual synthesis reaction in protein pocket and one-step retrosynthetic reagent replacement. Not only can growth be carried out at the reaction warhead, but also partial substructures in the molecule can be selected for retrosynthetic analysis and reagent replacement.

[0032] 4) In the process of molecular design, through virtual synthesis and interaction descriptor calculations, new ideas for drug design can be provided for systems with existing protein and small molecule complex structures, and new molecules with better protein-ligand interactions and easier synthesis can be discovered. Attached Figure Description

[0033] Figure 1 This is a flowchart of the present invention;

[0034] Figure 2 Create a flowchart for the synthesizer;

[0035] Figure 3 A schematic diagram of the synthesizer is prepared;

[0036] Figure 4 A schematic diagram illustrating the use of synthesizers;

[0037] Figure 5 Visualization of the results. Detailed Implementation

[0038] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0039] See Figure 1 A targeted drug design method based on chemical reactions and molecular building blocks, comprising the following specific steps:

[0040] 1) Commonly used reactions in pharmaceutical chemical synthesis were searched from the literature, and chemical reaction templates were extracted to construct a chemical reaction template database using SMIRDKS, totaling 56 entries;

[0041] 2) Based on the chemical reaction template database in step 1, extract the SMARTS of the reactant portion and construct a reactant chemical environment database and a synthon preparation reaction database using SMARTS representation;

[0042] 3) Select molecules with a heavy atom number of less than or equal to 20 from the Enamine molecular building block database, neutralize salt-containing molecules, and filter molecules containing elements other than C, H, O, N, S, P, Si, F, Cl, Br, I, and B, for a total of 2,404,960 molecules;

[0043] 4) Based on the synthons obtained in step 2, the molecular building blocks obtained in step 3 were converted into synthons using the RDKit. The reagent information and reaction type information were recorded, resulting in a total of 6,374,590 synthons. For details, please refer to the appendix. Figure 2 and attached Figure 3 ;

[0044] 5) Calculate the formal charge of the compound in step 4 and the protonated state of the compound in step 4 at pH=7;

[0045] 6) Use RDKit to generate initial three-dimensional conformations for the compound in step 5, up to 200 conformations;

[0046] 7) Based on the reactant chemical environment database and synthon preparation reaction database obtained in step 2, record the growth point atom ID, the ID of all atoms retained and the ID of all atoms deleted, the molecular building blocks SMILES and the reaction type of the three-dimensional molecules in step 6, and construct the synthon database.

[0047] 8) Based on the chemical environment database of reactants obtained in step 2, calculate the reaction warhead on the seed compound (or the smallest substructure obtained through the reaction in the database, which can be grown by one-step retrosynthesis), and record the retained part atom ID and deleted part atom ID on the seed compound;

[0048] 9) The distance between the protein and the seed compound The inner part is defined as a protein pocket, and spatial relative positional relationship descriptors and fragment global property descriptors are calculated;

[0049] 10) Based on the spatial relative position relationship descriptor and fragment global property descriptor obtained in step 9, search for a suitable synthesizer in the synthesizer database obtained in step 7.

[0050] 11) Based on the synthon and seed compound obtained in step 10, a series of new molecules that form specified interactions and meet the fragment requirements are obtained through virtual synthesis reactions at the growth point (see Appendix). Figure 4 The results will be visualized (see appendix). Figure 5 ).

[0051] This invention utilizes a constructed chemical reaction template database, a molecular building block database, and a molecular growth algorithm for targeted drug design. The molecules in the chemical space formed by the chemical reaction template database and the molecular building block database naturally possess high syntheticity. By virtually synthesizing seed compounds in a protein pocket, syntheticity is ensured while improving the binding ability of molecules, providing a new approach for targeted drug design.

[0052] Table 1. Chemical reaction templates (represented by SMIRKS) and chemical environments used in retrosynthetic analysis (represented by SMARTS)

[0053]

[0054]

[0055]

[0056]

[0057]

[0058] The above is merely a further description of the present invention and is not intended to limit the present invention. All equivalent implementations of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A targeted drug design method based on chemical reactions and molecular building blocks, characterized in that, The method specifically includes the following steps: Step 1: Search the literature for commonly used reactions in pharmaceutical chemical synthesis, extract chemical reaction templates, and construct a chemical reaction template database using SMIRKS; Step 2: Based on the chemical reaction template database in Step 1, extract the SMARTS of the reactant portion and construct a reactant chemical environment database and a synthon preparation reaction database using SMARTS representation; Step 3: Select molecules with 20 or fewer heavy atoms from the Enamine, BLDpharm, ChemDiv, WuxiAppTec, Chemieliva, StarshineChemical, Ambinter, and Mcule molecular building block databases; neutralize salt-containing molecules; and filter molecules containing elements other than C, H, O, N, S, P, Si, F, Cl, Br, I, and B. Step 4: Based on the synthon preparation reaction database obtained in Step 2, the molecular building blocks obtained in Step 3 are converted into synthons using RDKit, and the reaction reagent information and reaction type information are recorded; Step 5: Calculate the formal charge of the compound in Step 4 and the protonated state of the compound in Step 4 at pH=7; Step 6: Use RDKit to generate the initial three-dimensional conformation for the compound from Step 5; Step 7: Based on the reactant chemical environment database and synthon preparation reaction database obtained in Step 2, record the growth point atom ID, retained structure atom ID, deleted structure atom ID, molecular building blocks SMILES and reaction type of the three-dimensional molecules in Step 6, and construct the synthon database; Step 8: Based on the reactant chemical environment database obtained in Step 2, calculate the reaction warhead on the seed compound or the smallest substructure obtained through the reaction in the database, i.e., the fragment growth of reagent replacement by one-step retrosynthetic analysis, and record the retained part atom ID and deleted part atom ID on the seed compound; Step 9: Define the portion of the protein within 9 Å of the seed compound as a protein pocket, and calculate the spatial relative position descriptor and the fragment global property descriptor; Step 10: Based on the spatial relative position relationship descriptor and fragment global property descriptor obtained in Step 9, search for a suitable synthesizer in the synthesizer database obtained in Step 7; Step 11: Based on the synthon and seed compound obtained in Step 10, a series of new molecules that form specified interactions and meet the fragment requirements are obtained through virtual synthesis reaction at the growth point, and the results are visualized.

2. The targeted drug design method according to claim 1, characterized in that, The construction of the chemical reaction template database using SMIRKS described in step 1 is constructed by using SMARTS to represent the chemical environment constraints of reactants and products.

3. The targeted drug design method according to claim 1, characterized in that, The construction of the reactant chemical environment database and synthon preparation reaction database using SMARTS in step 2 is based on the constraints of the chemical environment during the breaking of old bonds and the formation of new bonds in the reaction template represented by SMARTS, as well as the smallest fragments newly generated in the reaction.

4. The targeted drug design method according to claim 1, characterized in that, The conversion of molecular building blocks into synthons in step 4 is based on the conversion of molecular building blocks into synthons by the reaction module in RDKit.

5. The targeted drug design method according to claim 1, characterized in that, The growth point atom ID mentioned in step 7 is obtained by substructure matching based on RDKit, which yields the corresponding growth point atom ID, the IDs of all atoms that are retained, and the IDs of all atoms that are deleted.

6. The targeted drug design method according to claim 1, characterized in that, Step 8, which involves calculating the reaction warhead on the seed compound or the smallest substructure obtained through the reaction in the database (i.e., fragment growth via reagent replacement through one-step retrosynthetic analysis), is based on SMARTS in the chemical template database of Step 1 and the reactant chemical environment database of Step 2, using RDKit. Specifically, this includes: i) Extract the SMARTS of reactants from the reactant chemical environment database and use RDKit to find matching reaction warheads on seed compounds; ii) Extract the smallest newly generated reaction unit from each reaction template in the chemical reaction template database, represent it in the form of SMARTS, find matching substructures on the seed compound using RDKit, and perform molecular growth by replacing one of the reagents in the reaction.

7. The targeted drug design method according to claim 1, characterized in that, Step 11 specifically includes: superimposing the seed compound and the synthon together according to the recorded growth point atom IDs, then removing some atoms and connecting the remaining atoms to complete molecular growth; based on the positions of the synthon and the seed compound at the growth points obtained in step 10, a series of new molecules that form specified interactions and meet the fragment requirements are obtained through virtual synthesis reactions, and the results are visualized.

Images