High-throughput molecule screening and molecule intelligent construction platform and data processing method thereof

Abstract

The invention provides a high-throughput molecule screening and molecule intelligent construction platform and a data processing method thereof, and relates to the technical field of chemical informatics and drug research and development, the high-throughput molecule screening and molecule intelligent construction platform comprises a multi-molecule batch operation module, a single-molecule editing module and a modular task combination engine, 34 core functions of molecular data cleaning, multi-dimensional screening, molecular fingerprint similarity analysis, similarity map screening based on MACCS fingerprints, structure replacement, molecular disassembly, atomic contribution analysis and the like are supported; according to the data processing method, a unique SMILES code is generated through standardized cleaning, in combination with screening conditions such as user-defined functional groups and molecular weights, a similarity matrix is calculated by utilizing topology / MACCS fingerprints, and a reference molecular structure is matched through a similarity map, so that high-precision screening is realized. The problems of limited application scene, poor database compatibility and complex operation in the prior art are solved, and the molecular screening efficiency and precision are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of chemical informatics and drug development technology, and specifically to a high-throughput molecular screening and molecular intelligence platform and a data processing method thereof. Background Art

[0002] Commercial molecular simulation software, such as the Schrödinger Suite, MOE (Molecular Operating Environment), and OpenEye Scientific Software, primarily rely on molecular docking techniques to rapidly predict target protein-ligand binding patterns. Core functions include optimizing protein-ligand complex structures using Lamarckian genetic algorithms or Monte Carlo simulations, virtual screening of ligand databases based on 3D shape similarity and pharmacophore characteristics, and providing molecular visualization tools (such as MOE's sequence editor and Schrödinger's Maestro interface) for real-time molecular structure modeling and parameter adjustment.

[0003] The open-source tool RDKIT is primarily developed in the Python programming language. Developers can use programming to efficiently and accurately perform a variety of complex molecular operations, such as chemical information processing, molecular fingerprint comparison, and molecular editing.

[0004] In the research and development process of medicine and chemical engineering, molecular screening and molecular editing are key links. Efficient and accurate molecular screening can reduce experimental costs and shorten experimental cycles. Current screening tools are mostly specialized applications for drug development and have the following main shortcomings:

[0005] (1) Limitations in applicable screening scenarios: Molecular docking-based screening technologies primarily focus on the pharmacodynamic characteristics of molecules and their interactions with proteins. This type of screening model, which relies on specific algorithms, is difficult to apply to scenarios that focus on molecular performance and where no protein is present.

[0006] (2) Poor database flexibility and compatibility: In this type of software, the database is generally a built-in database, and it is impossible to switch to a more suitable special database for different screening occasions;

[0007] (3) High operational complexity: RDKIT requires mastery of programming language to use, and ordinary users cannot use it quickly. Summary of the Invention

[0008] In view of this, in order to solve the problems of current screening tools such as limited screening occasions, poor database flexibility and compatibility, and high operational complexity, the purpose of the present invention is to propose a high-throughput molecular screening and molecular intelligence platform and its data processing method, which adopts a customizable screening module, supports multi-angle screening condition configuration, and sets screening conditions according to the use occasion; adopts a screening method based on similarity map, starting from the perspectives of molecular structure, properties, etc., to screen potential molecules that are highly similar to the research object, and the screening results are more accurate and reliable; with a wide range of databases, users can download or import custom databases online from multiple angles such as atomic type, molecular weight, aromaticity, etc. for subsequent processing, so that the screening process has strong pertinence; it realizes complete visual operation, adopts modular architecture design, and the platform has comprehensive functions. It can realize a variety of combination schemes through process arrangement, which can meet basic learning and scientific research needs, effectively support scientific research scenarios such as drug molecule design and molecular screening, and is a powerful efficiency tool in theoretical learning and experimental design.

[0009] To achieve the above object, the present invention provides the following technical solutions:

[0010] Based on the above objectives, in a first aspect, the present invention provides a high-throughput molecular screening and molecular intelligence platform, comprising the following components:

[0011] Multi-molecule batch operation module, including:

[0012] Molecular cleaning unit: used to clean the input molecular data containing SMILES codes, standardize the SMILES codes, and filter out invalid and repeated SMILES codes;

[0013] Screening unit: supports user-defined screening conditions and performs multi-dimensional screening on the molecular database uploaded by users;

[0014] Molecular fingerprint similarity calculation unit: calculates the molecular fingerprint of each molecule in the molecular database, calculates the similarity between every two molecular fingerprints based on the molecular fingerprint, and stores the calculation results in an N×N similarity matrix;

[0015] Difference analysis unit: based on the similarity between molecular fingerprints, it selects the molecular sets whose differences exceed the threshold range;

[0016] Similarity map screening unit: establishes a similarity map between the reference molecule and each molecule in the molecule set through MACCS fingerprints and records the similarity between each molecule and the reference molecule, and screens out target molecules based on the similarity threshold;

[0017] Data conversion unit: based on the screened target molecules, batch convert them into corresponding MOL files according to the standardized SMILES encoding;

[0018] Single-molecule editing module, including:

[0019] Structural replacement unit: used to read the SMILES encoding of the molecule to be processed, the target structure, and the new structure, extract the target structural features, compare them with the various molecular fragments in the molecule to be processed, and replace the target substructure with the new structure by matching it. It supports automatic filtering of local replacements and unreasonable replacements;

[0020] Molecular disassembly unit: Use the Recap method or Brics method to perform retrosynthetic disassembly or bond synthesis disassembly on molecules to generate a tree data structure or a deduplication list;

[0021] Atomic Contribution Analysis Unit: Calculates the contribution of atoms to molecular properties based on electronegativity, atomic radius, and hydrogen bond donor / acceptor sites;

[0022] Modular task combination engine: allows users to arrange core functional modules as needed to complete at least one combination process among high-throughput screening, reverse synthesis analysis, and similar molecular screening.

[0023] In a further embodiment of the present invention, the high-throughput molecular screening and molecular intelligence platform further comprises:

[0024] Exogenous database interface: compatible with online import of user-defined molecular libraries, independent of the built-in database, and supports uploading and processing of CSV / SDF format files;

[0025] Visual interactive interface: provides real-time display of molecular structure, task parameter configuration, result visualization output, and drag-and-drop save functions, without the need for programming.

[0026] As a further embodiment of the present invention, the molecular cleaning unit performs cleaning processing on the SMILES-encoded molecular data, including dehydrogenation, removal of counterions, processing of tautomers, testing of molecular chirality, filtering out invalid molecules, and deleting duplicate molecules; dehydrogenation is performed on the molecular data and the SMILES encoding of the molecules is standardized, the main molecular fragments are retained and the counterions in the molecules are deleted, the tautomers are processed and the molecular chirality is tested, invalid molecules are filtered out, and duplicate molecules are deleted, so as to standardize the SMILES-encoded molecular data.

[0027] As a further solution of the present invention, in the screening unit, the user sets screening conditions from at least one of the following perspectives: functional group type, number of functional groups, molecular weight, and aromaticity; when performing multi-dimensional screening on the molecular database uploaded by the user, the SMILES code of the molecules in the original database uploaded by the user is read and compared with the set screening conditions to obtain molecular data that meets the requirements to complete the screening, and the molecular data output by the screening contains the corresponding screening information.

[0028] As a further embodiment of the present invention, in the molecular data output by the screening unit, the screening information includes at least one of the number of functional groups, the number of aromatic rings, the molecular weight, and the lipid-water partition coefficient (logP).

[0029] As a further solution of the present invention, the similarity map screening unit performs the following steps:

[0030] Generate MACCS fingerprint similarity maps between reference molecules and database molecules;

[0031] Automatically filter low-similarity molecules based on the user-set similarity threshold;

[0032] Outputs the filtered results including similarity scores.

[0033] As a further solution of the present invention, the multi-molecule batch operation module also includes a molecular side chain excision unit, which is used to input the CSV file to be processed and the column name of the column where the molecular SMILES code is located, input the SMILES code of the side chain structure in the reference structure input area, and perform side chain excision on the molecules in the CSV after selecting the task type.

[0034] As a further solution of the present invention, in the molecular disassembly unit, the tree data structure generated by the Recap method supports disassembly path backtracking, and the list generated by the Brics method contains a mapping relationship between atomic numbers and reaction types.

[0035] As a further solution of the present invention, the atomic contribution analysis unit is further used to mark the contribution size of the position atoms in the molecular 2D / 3D model through color coding.

[0036] As a further solution of the present invention, the modular task combination engine includes the following categories, which users can customize:

[0037] A combination of rapid screening tasks for candidate molecules, including: molecular cleaning, multi-condition screening, molecular property calculation, differential analysis, and MOL file generation;

[0038] A combination of retrosynthesis analysis tasks, including: molecular cleaning, molecular skeleton extraction, molecular disassembly, substructure positioning, and atomic contribution analysis;

[0039] Similar molecule screening task combination, including: molecule cleaning, multi-condition screening, similarity map screening and MOL file generation.

[0040] In a second aspect, the present invention further provides a data processing method for a high-throughput molecular screening and molecular intelligence platform, comprising the following steps:

[0041] For multi-molecule batch operations, it receives the molecular database files uploaded by users and cleans the molecular data containing SMILES encoding in the files;

[0042] Receive user-defined screening conditions, compare the standardized SMILES code with the screening conditions, output the set of molecules that meet the requirements, and record the screening index information in the results;

[0043] Calculate the similarity between every two molecules in the molecular set, generate an N×N similarity matrix, and filter out the molecular subsets whose differences exceed the threshold range based on the similarity matrix;

[0044] Receive the SMILES code of the reference molecule, generate the corresponding MACCS fingerprint, build a similarity map between the reference molecule and the database molecule, calculate the similarity score between each molecule and the reference molecule; filter out the target molecules according to the preset similarity threshold, and output the screening results including the similarity;

[0045] Convert the SMILES codes of the screened target molecules into standard MOL files in batches;

[0046] Perform at least one of the following actions for single molecule edits:

[0047] Structural replacement: match the target substructure and replace it with a new structure, automatically filtering unreasonable replacement sites;

[0048] Molecular disassembly: Use the Recap method to generate a tree structure of the retrosynthetic disassembly path, or use the Brics method to generate a deduplicated list of bond synthesis disassembly;

[0049] Atomic contribution analysis: Calculate atomic contributions based on electronegativity, atomic radius, and hydrogen bonding sites, and annotate them in the molecular model through color coding;

[0050] For different application scenarios, users can arrange task combinations as needed.

[0051] As a further solution of the present invention, the molecular data containing SMILES encoding in the file is cleaned, comprising the following steps:

[0052] Perform dehydrogenation and normalization operations on SMILES codes;

[0053] Remove the counterion from the molecule and retain the main fragment;

[0054] handle tautomers and examine molecular chirality;

[0055] Invalid molecules were filtered out and duplicates were deleted to generate a standardized and unique SMILES encoding database.

[0056] As a further embodiment of the present invention, user-defined screening conditions are received, including at least one of functional group type, number of functional groups, molecular weight, and aromaticity.

[0057] As a further solution of the present invention, users can arrange task combinations as needed for different application scenarios, including:

[0058] A combination of rapid screening tasks for candidate molecules, including: molecular cleaning, multi-condition screening, molecular property calculation, differential analysis, and MOL file generation;

[0059] A combination of retrosynthesis analysis tasks, including: molecular cleaning, molecular skeleton extraction, molecular disassembly, substructure positioning, and atomic contribution analysis;

[0060] Similar molecule screening task combination, including: molecule cleaning, multi-condition screening, similarity map screening and MOL file generation.

[0061] As a further embodiment of the present invention, the similarity between every two molecules in a molecular set is calculated based on a topological fingerprint or a MACCS fingerprint, wherein:

[0062] Used to accurately match molecular structures when calculated based on topological fingerprints and Tanimoto coefficients;

[0063] When calculated based on MACCS fingerprint and Dice coefficient, it is used for fast screening or substructure matching;

[0064] The diagonal elements in the generated N×N similarity matrix are set to zero, and the off-diagonal elements represent the similarity between molecules.

[0065] Compared with the existing technology, the high-throughput molecular screening and molecular intelligence platform and its data processing method proposed in this invention have achieved significant improvements in multi-dimensional screening accuracy, operational flexibility, data compatibility, task combination efficiency, and user experience through technological innovation and system architecture optimization. It has the following beneficial effects:

[0066] 1. Multi-dimensional configurable screening modules expand applicable scenarios and improve screening efficiency.

[0067] The present invention configures a multi-dimensional screening module to support users in setting complex screening conditions based on dimensions such as functional group type and number of functional groups, thereby achieving targeted optimization of specific application scenarios and enhancing scenario adaptability. It breaks through the limitation of traditional molecular docking technology's dependence on proteins and is suitable for molecular performance screening scenarios without protein participation. The data is pre-processed by a standardized cleaning module to improve screening efficiency. The output results contain detailed indicators of the matching of screening conditions, which can help users quickly locate key molecules.

[0068] 2. The screening model based on similarity map significantly improves the accuracy of molecular matching.

[0069] The present invention uses MACCS fingerprints to generate molecular structure and property similarity maps, and combines them with user-set similarity thresholds to screen target molecules. It has efficient and reliable similarity analysis technology. By introducing a screening model based on similarity maps, it deeply analyzes molecular structure, molecular properties and three-dimensional similarity, constructs a similarity map to screen molecules, and realizes a reliable screening model.

[0070] 3. It has a fully exogenous molecular database architecture and flexible data compatibility to improve experimental targeting.

[0071] The present invention adopts the design principle of zero built-in database and is compatible with the online import of user-defined molecular libraries to achieve targeted screening. It supports users to upload molecular libraries in CSV / SDF format and standardizes them through the cleaning module. Users can import special databases according to specific research needs to avoid interference from redundant data in general databases. The generated standardized MOL files can be directly used for molecular dynamics simulations or quantum chemistry calculations, reducing the data preprocessing steps.

[0072] 4. Visual interaction and modular task combination lower the operation threshold and enhance functional scalability.

[0073] The platform of the present invention also has a graphical human-computer operation interface, develops a visual interactive interface, supports drag-and-drop operations and visual export of results, reduces the difficulty of operation, and adopts a modular architecture design through highly customized task combination strategies. Each module can run independently or be freely combined, covering the full process requirements from high-throughput initial screening to single-molecule refined editing. It can enable each core functional module to run independently and realize a variety of combination schemes through process orchestration, which can meet basic learning and scientific research needs and effectively support the use requirements of scientific research scenarios such as drug molecule design and molecular screening.

[0074] These and other aspects of the present application will be more clearly understood in the following description of the embodiments. It should be understood that the above general description and the following detailed description are merely exemplary and explanatory and cannot limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the following briefly introduces the drawings required for the exemplary embodiments or related technical descriptions. The drawings are used to provide a further understanding of the present invention and constitute part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the drawings:

[0076] Figure 12 is a structural block diagram of a high-throughput molecular screening and molecular intelligence platform according to an embodiment of the present invention.

[0077] Figure 2 Schematic diagram of data comparison before and after cleaning molecules in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0078] Figure 3 Schematic diagram of the comparison of data before and after screening in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0079] Figure 4 This is a display diagram of post-screening data in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0080] Figure 5 This is a graph of the output results of the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention, which calculates the similarity (topological fingerprint and Tanimoto) from molecular fingerprints.

[0081] Figure 6 This is a graph showing the output results of the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention, which calculates similarity (MACCS fingerprint and Dice) based on molecular fingerprints.

[0082] Figure 7 This is a schematic diagram of the output results of molecules with large differences based on molecular fingerprint search in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0083] Figure 8 This is a schematic diagram of the output results of molecules with smaller differences based on molecular fingerprint search in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0084] Figure 9 This is a partial schematic diagram of the output results of molecules with smaller differences based on molecular fingerprint search in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0085] Figure 10 This is a graph of the output results of screening molecules based on a similarity map in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0086] Figure 11 This is a diagram showing the output results of the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0087] Figure 12 This is a diagram of the structural replacement output results in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0088] Figure 13 This is a diagram of the molecular recap output results in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0089] Figure 14 This is a diagram of the output results of molecular disassembly (BRICS) in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0090] Figure 15 This is a graph showing the output results of calculating the atomic contribution in a molecule in the high-throughput molecular screening and molecular intelligence platform according to an embodiment of the present invention.

[0091] Figure 16 Schematic diagram of the input area in the high-throughput molecular screening and molecular intelligence platform according to an embodiment of the present invention.

[0092] Figure 17 This is a pop-up window showing the calculation of molecules in the high-throughput molecular screening and molecular intelligence platform according to an embodiment of the present invention.

[0093] Figure 18 This is a diagram for filling in relevant columns in the calculated molecules in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0094] Figure 19 This is a diagram for filling in the reference structure of calculated molecules in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0095] Figure 20 This is a task selection diagram for batch operations on multiple molecules in a high-throughput molecular screening and molecular intelligence platform according to an embodiment of the present invention.

[0096] Figure 21 Schematic diagram of executing tasks in computing molecules in the high-throughput molecular screening and molecular intelligence platform according to an embodiment of the present invention.

[0097] Figure 22 This is a SMILES input diagram for calculating molecules in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0098] Figure 23 This is the SMILES encoding input diagram for calculating the reference structure of molecules in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0099] Figure 24 This is a SMILES encoding input diagram for calculating replacement substructures in molecules in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention.

[0100] Figure 25 This is a task selection diagram for single-molecule editing in computational molecules in the high-throughput molecular screening and molecular intelligence platform according to an embodiment of the present invention.

[0101] Figure 26This is an error prompt and task progress chart for the output area of ​​the calculated molecules in the high-throughput molecular screening and molecular intelligence platform of an embodiment of the present invention. DETAILED DESCRIPTION

[0102] Below, the present application is further described in conjunction with the accompanying drawings and specific implementation methods. It should be noted that, under the premise of no conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0103] To make the purpose, technical solutions and advantages of the present invention more clearly understood, the following is a further detailed description of the embodiments of the present invention in conjunction with specific embodiments and with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application.

[0104] It should be noted that all expressions using "first" and "second" in the embodiments of the present invention are intended to distinguish two non-identical entities or non-identical parameters with the same name. Therefore, "first" and "second" are used for convenience of expression only and should not be understood as limitations on the embodiments of the present invention. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, other steps or units inherent to a process, method, system, product, or device that includes a series of steps or units.

[0105] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0106] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, combined, or partially merged, so the actual execution order may vary depending on the actual situation.

[0107] The following describes some embodiments of the present application in detail with reference to the accompanying drawings. In the absence of conflict, the following embodiments and features therein may be combined with each other.

[0108] To address the limitations of current screening tools, poor database flexibility and compatibility, and high operational complexity, this paper proposes a high-throughput molecular screening and molecular intelligence platform and its data processing method. Compared with traditional screening software, this platform offers significant advantages and overcomes these deficiencies:

[0109] (1) Customizable multi-angle screening module: The present invention adopts a customizable screening module that supports multi-angle screening condition configuration (including functional group type, number of functional groups, molecular weight, etc.) and sets the screening conditions according to the application scenario.

[0110] (2) Screening method based on similarity map: The present invention adopts a screening method based on similarity map, starting from the perspective of molecular structure, properties, etc., to screen potential molecules that are highly similar to the research object, and the screening results are more accurate and reliable.

[0111] (3) Extensive Database: This invention only processes the molecular database uploaded by users and does not add any built-in database. Users can download or import their own customized databases online for subsequent processing from multiple perspectives such as atomic species, molecular weight, and aromaticity, making the screening process more targeted.

[0112] (4) Complete visualization and simple operation: The present invention provides a relatively complete operation interface, which allows users to operate it proficiently without having to master programming languages. At the same time, the results are output visually, making information processing more efficient.

[0113] (5) Complete functions and high combination degree: The platform developed by the present invention is comprehensive in functions and adopts a modular architecture design. The 34 core functional modules can operate independently and realize various combination schemes through process arrangement.

[0114] The high-throughput molecular screening and molecular intelligence platform and its data processing method of the present invention cover a variety of core functional modules, including two major task types: multi-molecule batch operation and single-molecule editing. The processing logic is: input the file to be processed, then input the column name where the SMILES is located, or input the molecule to be processed, select the task type, if the task requires defining a reference structure, then determine whether the reference structure has been defined, if not defined, issue a prompt, if it has been defined, execute the task, if there is no exception, output the processing result, if there is an exception, output the exception information. Among them, at the multi-molecule batch operation level, the application supports large-scale molecular screening, molecular cleaning, molecular property calculation, etc., and can complete the data processing of thousands of molecules at a time; in the single-molecule editing module, users can perform refined operations such as structure replacement, conformation and force field optimization. The platform is comprehensive and sophisticated. While each function is independent of each other, it can adapt to different usage scenarios through the combination of tasks. In addition, users can perform diversified operations through a visual interface, improving the efficiency of data processing. The platform can meet basic learning and scientific research needs, effectively supporting scientific research scenarios such as drug molecule design and molecular screening, and is a powerful efficiency-enhancing tool in theoretical learning and experimental design.

[0115] The high-throughput molecular screening and molecular intelligence platform and its data processing method provided by the present invention adopt a customizable screening module, support multi-angle screening condition configuration, and set screening conditions according to the usage scenario; adopt a screening method based on similarity map, starting from the perspectives of molecular structure, properties, etc., to screen potential molecules that are highly similar to the research object, and the screening results are more accurate and reliable; a wide range of databases, when used, users can download or import custom databases online from multiple perspectives such as atomic type, molecular weight, aromaticity, etc. for subsequent processing, making the screening process more targeted; realize complete visual operation, adopt a modular architecture design, the platform has comprehensive functions, and can realize various combination schemes through process orchestration, which can meet basic learning and scientific research needs, effectively support scientific research scenarios such as drug molecule design and molecular screening, and is a powerful efficiency tool in theoretical learning and experimental design.

[0116] See also Figure 1 As shown, an embodiment of the present invention provides a high-throughput molecular screening and molecular intelligence platform, including a multi-molecule batch operation module and a single-molecule editing module, wherein the multi-molecule batch operation module includes:

[0117] Molecular cleaning unit: used to clean the input molecular data containing SMILES codes, standardize the SMILES codes, and filter out invalid and repeated SMILES codes;

[0118] Screening unit: supports user-defined screening conditions and performs multi-dimensional screening on the molecular database uploaded by users;

[0119] Molecular fingerprint similarity calculation unit: calculates the molecular fingerprint of each molecule in the molecular database, calculates the similarity between every two molecular fingerprints based on the molecular fingerprint, and stores the calculation results in an N×N similarity matrix;

[0120] Difference analysis unit: based on the similarity between molecular fingerprints, it selects the molecular sets whose differences exceed the threshold range;

[0121] Similarity map screening unit: establishes a similarity map between the reference molecule and each molecule in the molecule set through MACCS fingerprints and records the similarity between each molecule and the reference molecule, and screens out target molecules based on the similarity threshold;

[0122] Data conversion unit: Based on the screened target molecules, batch convert them into corresponding MOL files according to the standardized SMILES encoding.

[0123] In this embodiment, see Figure 2As shown in the figure, the molecular cleaning unit cleans the SMILES-encoded molecular data by dehydrogenation, removing counterions, processing tautomers, checking molecular chirality, filtering out invalid molecules, and deleting duplicate molecules. The molecular data is dehydrogenated and the SMILES encoding of the molecules is standardized. The main molecular fragments are retained and the counterions in the molecules are deleted. Tautomers are processed and the molecular chirality is checked. In addition, invalid molecules are filtered out and duplicate molecules are deleted. This standardizes the SMILES-encoded molecular data and ensures the uniqueness of each molecule in the database. Figure 2 As shown, the SMILES within the rectangular wireframe encode different expressions of the same molecule, and duplicates have been effectively removed after cleaning.

[0124] See also Figure 3 As shown, in the screening unit, the user sets screening conditions from at least one of the following perspectives: functional group type, number of functional groups, molecular weight, and aromaticity; when performing multi-dimensional screening on the molecular database uploaded by the user, the SMILES code of the molecules in the original database uploaded by the user is read and compared with the set screening conditions to obtain molecular data that meets the requirements to complete the screening, and the molecular data output by the screening contains the corresponding screening information. Among them, the screening information in the molecular data output by the screening unit includes at least one of the following: number of functional groups, number of aromatic rings, molecular weight, and lipid-water partition coefficient. For example, see Figure 4 As shown, the setting is to filter out molecules with less than 3 aromatic rings, and a column recording the number of aromatic rings contained in the molecule will be added to the output results.

[0125] In this embodiment, when the molecular fingerprint similarity calculation unit calculates the molecular fingerprint of each molecule in the molecular database, two methods are used to calculate similarity based on the molecular fingerprint: a calculation method based on topological fingerprint and Tanimoto, and a calculation method based on MACCS fingerprint and Dice. Compared with the calculation method based on topological fingerprint and Tanimoto, the calculation method based on MACCS fingerprint and Dice is more coarse and suitable for situations requiring rapid screening or substructure matching.

[0126] When calculating molecular fingerprint similarity, first calculate the molecular fingerprint of each molecule in the database, then calculate the similarity between each two molecular fingerprints, and store the calculation results in an N×N similarity matrix. Figure 5 and Figure 6 As shown in the output matrix, the diagonal positions represent the similarity between each molecule and itself. To reduce computational complexity, these values ​​are not analyzed and are defaulted to 0. A similarity of 1 indicates that the two molecules are completely identical. Note that in this matrix, the data symmetric about the diagonal represents the same meaning. For example, the similarities S_ij and S_ji correspond to the same meaning, representing the similarity between molecule number i and molecule number j in the database.

[0127] In the difference analysis unit, when searching for molecules with large / small differences based on molecular fingerprints, MACCS fingerprints are used to search for molecules with large differences in the file. This function is suitable for occasions with large amounts of data. It can quickly screen out molecules with obvious differences / similarities as research objects to reduce preparatory time and experimental costs. First, the molecular fingerprints of each molecule in the database are calculated, and then the similarity between each two molecular fingerprints is calculated. Based on the similarity, molecules with large / small differences are obtained, such as Figure 7 and Figure 8 shown.

[0128] Figure 9 As shown, the marked positions are the differences between the two molecules. It can be seen that the results of searching for molecules with smaller differences based on molecular fingerprints are very reliable.

[0129] In this embodiment, the similarity map screening unit performs the following steps:

[0130] Generate MACCS fingerprint similarity maps between reference molecules and database molecules;

[0131] Automatically filter low-similarity molecules based on the user-set similarity threshold;

[0132] Outputs the filtered results including similarity scores.

[0133] Among them, the similarity map is an image drawn based on the molecular fingerprint, which can reflect the various characteristics of the molecule in structure and properties. This function uses the MACCS fingerprint to establish a similarity map between the reference molecule and each molecule in the file, and records the similarity between each molecule and the reference molecule based on the similarity map. By setting the similarity threshold, molecules with low similarity are filtered out, thereby screening out molecules with structures and properties that are relatively similar to the reference molecule. This function is suitable for screening similar molecules of the same type based on existing functional molecules, such as Figure 10 As shown, it can be seen that the similarity between the screened molecules and the reference molecules is low, indicating that the reference molecules are quite different from the molecules in the file in terms of structure and properties.

[0134] In this embodiment, when the data conversion unit generates the corresponding MOL file from the SMILES in the table, in the intersection of chemical informatics and drug development, the SMILES code in the database is efficiently converted into a standard MOL file using automated technology. Before performing this task, the data in the database needs to be cleaned up molecularly to ensure the standardization of the SMILES code. Furthermore, by interpreting the SMILES code of each molecule, the corresponding MOL file is generated for subsequent operations, such as Figure 11 As shown, it should be noted that the MOL file generated here has not been structurally optimized.

[0135] Among them, the multi-molecule batch operation module also includes a molecular side chain excision unit, which is used to input the CSV file to be processed and the column name of the column where the molecular SMILES code is located, input the SMILES code of the side chain structure in the reference structure input area, and perform side chain excision on the molecules in the CSV after selecting the task type.

[0136] In this embodiment, the single molecule editing module comprises:

[0137] Structural replacement unit: used to read the SMILES encoding of the molecule to be processed, the target structure, and the new structure, extract the target structural features, compare them with the various molecular fragments in the molecule to be processed, and replace the target substructure with the new structure by matching it. It supports automatic filtering of local replacements and unreasonable replacements;

[0138] Molecular disassembly unit: Use the Recap method or Brics method to perform retrosynthetic disassembly or bond synthesis disassembly on molecules to generate a tree data structure or a deduplication list;

[0139] Atomic Contribution Analysis Unit: Calculates the contribution of atoms to molecular properties based on electronegativity, atomic radius, and hydrogen bond donor / acceptor sites.

[0140] In this embodiment, structure replacement is a basic step in single molecule editing. First, the SMILES encoding of the molecule to be processed, the target structure, and the new structure are read, and the target structure features are further extracted and compared with the molecular fragments in the molecule to be processed. After a successful match, all target structures that meet the requirements are replaced with the new structure, such as Figure 12 As shown, please note that when the molecular structure after the local position structure replacement is unreasonable, the replacement operation will not be performed at that position, and other positions will be replaced normally.

[0141] In the molecular disassembly unit, the tree data structure generated by the Recap method supports disassembly path backtracking, and the list generated by the Brics method contains the mapping relationship between the atomic number and the reaction type. The atomic contribution analysis unit is also used to mark the contribution size of the position atoms in the molecular 2D / 3D model through color coding. Among them, the Brics method focuses more on whether the bond can be synthesized to perform disassembly, and returns a deduplicated list. The serial number on the atom corresponds to a specific reaction type. The disassembly result is as follows: Figure 13 and Figure 14 shown.

[0142] In the atomic contribution analysis unit, when calculating the contribution of atoms in a molecule, in the study of molecular property deconstruction, the atomic contribution will be analyzed from the following aspects: based on electronegativity, the polarization ability of atoms to the electron cloud distribution is evaluated; combining the difference between covalent radius and van der Waals radius, the dual characteristics of electron attraction and spatial shielding of atoms in the molecular skeleton are obtained; and hydrogen bond donors / acceptors or nucleophilic / electrophilic sites are located. It accurately reflects the characteristic structure of the molecule and provides a basis for predicting molecular properties. The output results are as follows: Figure 15 As shown, for example, the blue area can be used to indicate that the contribution of the atom at this position is large, and the atom is an important factor affecting the properties of the molecule; the red part indicates that the contribution of the atom at this position is small.

[0143] In this embodiment, the high-throughput molecular screening and molecular intelligence platform further includes:

[0144] Modular task combination engine: allows users to arrange core functional modules as needed to complete at least one combination of high-throughput screening, retrosynthesis analysis, and similar molecular screening;

[0145] Exogenous database interface: compatible with online import of user-defined molecular libraries, independent of the built-in database, and supports uploading and processing of CSV / SDF format files;

[0146] Visual interactive interface: provides real-time display of molecular structure, task parameter configuration, result visualization output, and drag-and-drop save functions, without the need for programming.

[0147] The modular task combination engine includes the following categories, which users can customize:

[0148] A combination of rapid screening tasks for candidate molecules, including: molecular cleaning, multi-condition screening, molecular property calculation, differential analysis, and MOL file generation;

[0149] A combination of retrosynthesis analysis tasks, including: molecular cleaning, molecular skeleton extraction, molecular disassembly, substructure positioning, and atomic contribution analysis;

[0150] Similar molecule screening task combination, including: molecule cleaning, multi-condition screening, similarity map screening and MOL file generation.

[0151] For example, the following are examples of common combination strategies, which users can flexibly modify or customize based on application requirements.

[0152] (1) Rapid screening combination of candidate molecules (applicable scenario: high-throughput molecular screening), the combination steps are:

[0153] Step 1: Get the original dataset;

[0154] Step 2: Extract the required columns from the table → obtain key information and reduce the amount of data;

[0155] Step 3: Clean the molecules → remove invalid SMILES and duplicates;

[0156] Step 4: Run the screening module → obtain molecules that meet the requirements;

[0157] Step 5: Run the calculation of molecular properties → select molecules with logP and MW in the specified range;

[0158] Step 6: Use molecular fingerprints to find molecules with large differences → obtain structural diversity samples;

[0159] Step 7: Batch generate corresponding MOL files for the SMILES in the table → obtain the MOL files of each molecule.

[0160] (2) Synthesis route reverse analysis and combination (applicable scenario: organic synthesis route reverse engineering), the combination steps are:

[0161] Step 1: Get the original dataset;

[0162] Step 2: Clean the molecules → remove invalid SMILES and duplicates;

[0163] Step 3: Extract molecular skeleton → identify core structure;

[0164] Step 4: Use molecular disassembly to analyze possible synthetic pathways;

[0165] Step 5: Substructure search → locate characteristic functional groups;

[0166] Step 6: Calculate atomic contributions → locate active sites.

[0167] (3) Similar molecule screening combination (applicable scenario: searching for potential molecules with existing reference molecules), the combination steps are:

[0168] Step 1: Get the original dataset;

[0169] Step 2: Extract the required columns from the table → obtain key information and reduce the amount of data;

[0170] Step 3: Clean the molecules → remove invalid SMILES and duplicates;

[0171] Step 4: Run the screening module → obtain molecules that meet the requirements;

[0172] Step 5: Run the similarity map to screen molecules → obtain molecules with similar performance to the reference molecules;

[0173] Step 6: Batch generate corresponding MOL files from the SMILES in the table → obtain the MOL files of each molecule.

[0174] The present invention configures a multi-dimensional screening module, allowing users to set complex screening conditions based on dimensions such as functional group type and number, enabling targeted optimization for specific application scenarios and enhancing adaptability. This overcomes the limitations of traditional molecular docking technology's reliance on proteins, making it suitable for molecular performance screening scenarios without protein involvement. Preprocessing data through a standardized cleaning module improves screening efficiency, and the output results include detailed indicators of matching screening conditions, helping users quickly locate key molecules. The present invention utilizes MACCS fingerprints to generate molecular structure and property similarity maps, combining them with user-set similarity thresholds to screen target molecules. This technology utilizes efficient and reliable similarity analysis technology. By introducing a screening model based on similarity maps, it constructs a similarity map to screen molecules through in-depth analysis of molecular structure, molecular properties, and three-dimensional similarity, thus achieving a reliable screening model.

[0175] The present invention adopts the design principle of zero built-in database and is compatible with the online import of user-defined molecular libraries to achieve targeted screening. It supports users to upload molecular libraries in CSV / SDF format and standardizes them through the cleaning module. Users can import special databases according to specific research needs to avoid interference from redundant data in general databases. The generated standardized MOL files can be directly used for molecular dynamics simulations or quantum chemistry calculations, reducing the data preprocessing steps.

[0176] The platform of the present invention also has a graphical human-computer operation interface, develops a visual interactive interface, supports drag-and-drop operations and visual export of results, reduces the difficulty of operation, and adopts a modular architecture design through highly customized task combination strategies. Each module can run independently or be freely combined, covering the full process requirements from high-throughput initial screening to single-molecule refined editing. It can enable each core functional module to run independently and realize a variety of combination schemes through process orchestration, which can meet basic learning and scientific research needs and effectively support the use requirements of scientific research scenarios such as drug molecule design and molecular screening.

[0177] In some embodiments, the high-throughput molecular screening and molecular intelligence platform of the present invention can also be developed and designed into a Web application of a molecular computing and data processing platform designed specifically for college students and scientific researchers in the fields of medicine and chemical engineering, covering 34 core functional modules, including two major task types: multi-molecule batch operation and single-molecule editing. Among them, at the multi-molecule batch operation level, the application supports large-scale molecular screening, molecular cleaning, molecular property calculation, etc., and can complete data processing of thousands of molecules at one time; in the single-molecule editing module, users can implement refined operations such as structure replacement, conformation and force field optimization. The application is comprehensive and sophisticated. While each function is independent of each other, it can adapt to different usage scenarios through the combination of tasks. In addition, users can implement diversified operations through a visual interface, which improves the efficiency of data processing. The application can meet basic learning and scientific research needs, effectively supports scientific research scenarios such as drug molecule design and molecular screening, and is a powerful synergistic tool in theoretical learning and experimental design.

[0178] The input area of ​​the web application can be divided into two categories: the upper part of the interface is the input area for batch operation of multiple molecules (such as set in the red box), and the lower part of the interface is the input area for single molecule editing (such as set in the blue box). Figure 16 As shown. The input area for multi-molecule batch operation includes the input file area, column name input area, reference structure input area, and task selection area; the input area for single molecule editing includes the SMILES input area for the molecule, the reference structure input area, the input area for replacing substructures, and the task selection area. When the mouse hovers over each input item, a prompt will pop up as shown below. Figure 17 In the Input File area, you can upload files to be processed. Supported file types include CSV and SDF. After selecting the file object, click Open to complete the upload.

[0179] In the column name input area, you need to enter the column name that records the molecular SMILES encoding in the CSV file for subsequent processing, such as Figure 18 As shown, for SDF files, this option does not need to be filled in. The reference structure input area is used to input the SMILES code of the reference structure, such as Figure 19 For example, when executing the task "Display multiple molecules based on a common skeleton", you need to enter the SMILES code of the common skeleton; in the task "Filter molecules based on similarity map", you need to enter the SMILES code of the reference molecule. Task 1 corresponds to the task option of multi-molecule batch operation. You can select the target task by clicking the drop-down arrow, such as Figure 20 After completing the above operations, click the execute button to run the task, as shown in the following example: Figure 21 As shown. The SMILES input area is used to input the SMILES code of a single molecule to be operated, such as Figure 22The reference structure input area is used to input the SMILES code of the target structure, such as Figure 23 As shown in the figure, for example, in the substructure search task, the substructure to be searched is input; in the structure replacement task, the substructure to be replaced is input. The replacement substructure input area is used to input the SMILES code of the new substructure in the structure replacement task, such as Figure 24 Task 2 corresponds to the task option of single molecule editing. You can select the target task by clicking the drop-down arrow, such as Figure 25 shown.

[0180] In this web application, the input and output area is located at the bottom of the application interface and includes error prompts, task progress, picture display, etc. Figure 26 As shown, missing required fields will generate an error message in the corresponding area, allowing users to easily adjust the settings. When a task includes image output, it will be displayed in the image display area for users to preview. Furthermore, for user convenience, dragging an image to a designated folder will save the original image. When a task includes file output, a folder named "result" will automatically be generated on the desktop to store the output file.

[0181] To prevent user errors and crashes, this web application incorporates an interlocking mechanism within the input area. Initially, only the "File Upload" and "SMILES Input" areas are open. After a file is uploaded, the single-molecule editing area is disabled, while the multi-molecule batch operation area is enabled. After SMILES input, the multi-molecule batch operation area is disabled, while the single-molecule editing area is enabled. Clicking Refresh restores the initial state.

[0182] The above are exemplary embodiments disclosed in the present invention, but it should be noted that various changes and modifications may be made without departing from the scope of the embodiments disclosed in the claims. The functions, steps and / or actions of the method claims according to the disclosed embodiments described herein do not need to be performed in any particular order. In addition, although the elements disclosed in the embodiments of the present invention may be described or required in individual form, they may also be understood as multiple unless expressly limited to the singular.

[0183] It should be understood that, as used herein, the singular form "a" or "an" is intended to include the plural form as well, unless the context clearly supports an exception. It should also be understood that, as used herein, "and / or" refers to any and all possible combinations of one or more of the items listed in association. The serial numbers of the embodiments disclosed in the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0184] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to limit the scope of the disclosure of the present invention (including the claims) to these examples. Within the spirit of the present invention, the technical features of the above embodiments or different embodiments may be combined, and many other variations exist in different aspects of the above embodiments, which are not provided in detail for the sake of clarity. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A high-throughput molecular screening and molecular intelligence platform, characterized by: Includes a multi-molecule batch operation module, a single-molecule editing module, and a modular task combination engine; The multi-molecule batch operation module comprises: Molecular cleaning unit: used to clean the input molecular data containing SMILES codes, standardize the SMILES codes, and filter out invalid and repeated SMILES codes; Screening unit: supports user-defined screening conditions and performs multi-dimensional screening on the molecular database uploaded by users; Molecular fingerprint similarity calculation unit: calculates the molecular fingerprint of each molecule in the molecular database, calculates the similarity between every two molecular fingerprints based on the molecular fingerprint, and stores the calculation results in an N×N similarity matrix; Difference analysis unit: based on the similarity between molecular fingerprints, it selects the molecular sets whose differences exceed the threshold range; Similarity map screening unit: establishes a similarity map between the reference molecule and each molecule in the molecule set through MACCS fingerprints and records the similarity between each molecule and the reference molecule, and screens out target molecules based on the similarity threshold; Data conversion unit: based on the screened target molecules, batch convert them into corresponding MOL files according to the standardized SMILES encoding; The single molecule editing module comprises: Structural replacement unit: used to read the SMILES encoding of the molecule to be processed, the target structure, and the new structure, extract the target structural features, compare them with the various molecular fragments in the molecule to be processed, and replace the target substructure with the new structure by matching it. It supports automatic filtering of local replacements and unreasonable replacements; Molecular disassembly unit: Use the Recap method or Brics method to perform retrosynthetic disassembly or bond synthesis disassembly on molecules to generate a tree data structure or a deduplication list; Atomic Contribution Analysis Unit: Calculates the contribution of atoms to molecular properties based on electronegativity, atomic radius, and hydrogen bond donor / acceptor sites; The modular task combination engine allows users to arrange core functional modules as needed to complete at least one combination process among high-throughput screening, reverse synthesis analysis, and similar molecule screening.

2. The high-throughput molecular screening and molecular intelligence platform according to claim 1, characterized in that: The high-throughput molecular screening and molecular intelligence platform also includes: Exogenous database interface: compatible with online import of user-defined molecular libraries, independent of the built-in database, and supports uploading and processing of CSV / SDF format files; Visual interactive interface: provides real-time display of molecular structure, task parameter configuration, result visualization output, and drag-and-drop save functions, without the need for programming.

3. The high-throughput molecular screening and molecular intelligence platform according to claim 1, characterized in that: The molecular cleaning unit performs cleaning processing on the SMILES-encoded molecular data, including dehydrogenation, removal of counterions, processing of tautomers, testing of molecular chirality, filtering out invalid molecules, and deleting duplicate molecules; dehydrogenation processing is performed on the molecular data and the SMILES encoding of the molecules is standardized, the main molecular fragments are retained and the counterions in the molecules are deleted, the tautomers are processed and the molecular chirality is tested, invalid molecules are filtered out, and duplicate molecules are deleted, so as to standardize the SMILES-encoded molecular data.

4. The high-throughput molecular screening and molecular intelligence platform according to claim 3, characterized in that: In the screening unit, the user sets screening conditions from at least one of the following perspectives: functional group type, number of functional groups, molecular weight, and aromaticity. When performing multi-dimensional screening on the molecular database uploaded by the user, the SMILES code of the molecules in the original database uploaded by the user is read and compared with the set screening conditions to obtain molecular data that meets the requirements to complete the screening. The molecular data output by the screening contains the corresponding screening information.

5. The high-throughput molecular screening and molecular intelligence platform according to claim 4, characterized in that: In the molecular data output by the screening unit, the screening information includes at least one of the number of functional groups, the number of aromatic rings, the molecular weight, and the lipid-water partition coefficient.

6. The high-throughput molecular screening and molecular intelligence platform according to claim 1, characterized in that: The similarity map screening unit performs the following steps: Generate MACCS fingerprint similarity maps between reference molecules and database molecules; Automatically filter low-similarity molecules based on the user-set similarity threshold; Outputs the filtered results including similarity scores.

7. The high-throughput molecular screening and molecular intelligence platform according to claim 1, characterized in that: The multi-molecule batch operation module also includes a molecular side chain excision unit, which is used to input the CSV file to be processed and the column name of the column where the molecular SMILES code is located, input the SMILES code of the side chain structure in the reference structure input area, and perform side chain excision on the molecules in the CSV after selecting the task type.

8. The high-throughput molecular screening and molecular intelligence platform according to claim 7, characterized in that: In the molecular disassembly unit, the tree data structure generated by the Recap method supports disassembly path backtracking, and the list generated by the Brics method contains the mapping relationship between atomic numbers and reaction types.

9. The high-throughput molecular screening and molecular intelligence platform according to claim 1, characterized in that: The modular task combination engine includes the following categories, which users can customize: A combination of rapid screening tasks for candidate molecules, including: molecular cleaning, multi-condition screening, molecular property calculation, differential analysis, and MOL file generation; A combination of retrosynthesis analysis tasks, including: molecular cleaning, molecular skeleton extraction, molecular disassembly, substructure positioning, and atomic contribution analysis; Similar molecule screening task combination, including: molecule cleaning, multi-condition screening, similarity map screening and MOL file generation.

10. A data processing method for a high-throughput molecular screening and molecular intelligence platform according to any one of claims 1 to 9, characterized in that: The following steps are involved: For multi-molecule batch operations, it receives the molecular database files uploaded by users and cleans the molecular data containing SMILES encoding in the files; Receive user-defined screening conditions, compare the standardized SMILES code with the screening conditions, output the set of molecules that meet the requirements, and record the screening index information in the results; Calculate the similarity between every two molecules in the molecular set, generate an N×N similarity matrix, and filter out the molecular subsets whose differences exceed the threshold range based on the similarity matrix; Receive the SMILES code of the reference molecule, generate the corresponding MACCS fingerprint, build a similarity map between the reference molecule and the database molecule, and calculate the similarity score between each molecule and the reference molecule; Filter out target molecules according to a preset similarity threshold and output the screening results including the similarity; Convert the SMILES codes of the screened target molecules into standard MOL files in batches; Perform at least one of the following actions for single molecule edits: Structural replacement: match the target substructure and replace it with a new structure, automatically filtering unreasonable replacement sites; Molecular disassembly: Use the Recap method to generate a tree structure of the retrosynthetic disassembly path, or use the Brics method to generate a deduplicated list of bond synthesis disassembly; Atomic contribution analysis: Calculate atomic contributions based on electronegativity, atomic radius, and hydrogen bonding sites, and annotate them in the molecular model through color coding; For different application scenarios, users can arrange task combinations as needed.

Images