Drug screening system and drug screening method

Through the drug screening system, the drug expression and sorting indicators are encoded, and the drug expression and sorting indicators are trained to generate candidate drugs, and the drug intensity is screened based on the sorting indicators, which solves the problem of drug development schedule and high funding in the existing technology, and realizes drug development Speed ​​up the process.

CN113053470BActive Publication Date: 2024-11-12IND TECH RES INST
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202010522168.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Priority Date
2020-03-16
Filing Date
2020-06-10
Publication Date
2024-11-12
Estimated Expiration
2040-06-10

AI Technical Summary

Technical Problem

The prior art is difficult to effectively reduce the time and funding of drug research and development, especially in the pilot compound optimization stage. The simulation software takes a long time and has a low success rate, making it difficult to cope with complex compound structures.

Method used

The drug screening system is adopted, including a coding module, a candidate drug generation module and a drug sorting module. By encoding drug expressions and sorting indicators, a generative adversarial network is trained to generate candidate drugs, and drug intensity is screened based on the generated drug sorting indicators.

Benefits of technology

Effectively reduce the timeline and funding of drug development, accelerate the development process, and improve the efficiency of drug screening.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113053470B_ABST
    Figure CN113053470B_ABST
Patent Text Reader

Abstract

A drug screening system includes an encoding module, a candidate drug generation module and a drug ranking module. The encoding module encodes a drug expression and at least one drug ranking index to generate a first encoding variable. The candidate drug generation module trains a generative adversarial network based on the first encoding variable to generate a plurality of candidate drugs, wherein each candidate drug has a generated drug expression and at least one generated drug ranking index. The drug ranking module sorts the strength of the candidate drugs based on at least one generated drug ranking index of each candidate drug to screen the drugs.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a drug screening system and a drug screening method. Background Art

[0002] The process from screening drugs to becoming truly useful new drugs may take a lot of time and money. Lead compound optimization is an important stage in the drug development process, and the average cost of the drug development process is 414 million US dollars (about 23% of the drug development process) and the research and development time of the drug development process is 2 years (about 15% of the drug development process). In the past, the optimization model using simulation software or expert experience has been difficult to cope with today's increasingly complex compound structures, because the operation of simulation software will take a lot of time and the success rate is low.

[0003] Therefore, how to effectively reduce the time and cost of drug development is an important issue at present. Summary of the invention

[0004] The present invention provides a drug screening system and a drug screening method, thereby effectively reducing the drug development time and cost.

[0005] The present invention provides a drug screening system, comprising an encoding module, a candidate drug generation module and a drug ranking module. The encoding module encodes a drug expression and at least one drug ranking index to generate a first encoding variable. The candidate drug generation module trains a generative adversarial network based on the first encoding variable to generate a plurality of candidate drugs, wherein each candidate drug has a generated drug expression and at least one generated drug ranking index. The drug ranking module sorts the strength of the candidate drugs based on at least one generated drug ranking index of each candidate drug to screen the drugs.

[0006] The present invention provides a drug screening method, comprising the following steps: encoding a drug expression and at least one drug ranking index to generate a first encoding variable. Based on the first encoding variable, training a generative adversarial network to generate a plurality of candidate drugs, wherein each candidate drug has a generative drug expression and at least one generative drug ranking index. Based on at least one generative drug ranking index of each candidate drug, the strength of the candidate drugs is ranked to screen the drugs.

[0007] The drug screening system and drug screening method disclosed in the present invention encode drug expressions and drug ranking indicators to generate a first encoding variable. Based on the first encoding variable, a generative adversarial network is trained to generate multiple candidate drugs, each of which has a generated drug expression and at least one generated drug ranking indicator. Based on the generated drug ranking indicator of each candidate drug, the strength of the candidate drugs is ranked to screen the drugs. In this way, the embodiment of the present invention can effectively reduce the time and cost of drug development and speed up the development process. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 is a schematic diagram of a drug screening system according to an embodiment of the present invention;

[0009] Figure 2 is a schematic diagram of coding variables according to an embodiment of the present invention;

[0010] Figure 3 is a schematic diagram of calculating drug structure distribution according to an embodiment of the present invention;

[0011] Figure 4 for Figure 1 Detailed schematic diagram of the candidate drug generation module;

[0012] Figure 5 is a flow chart of a drug screening method according to one embodiment of the present invention;

[0013] Figure 6 for Figure 5 Detailed flowchart of step S504.

[0014] [Description of Reference Numerals]

[0015] 100: Drug screening system

[0016] 110: Encoding module

[0017] 120: Drug candidate generation module

[0018] 130: Drug sorting module

[0019] 140: Display module

[0020] 210, 220: Area

[0021] 310: Compound

[0022] 410: Fabricating a drug generation module

[0023] 420: Judgment module

[0024] S502~S508, S602~S612: Steps DETAILED DESCRIPTION

[0025] The technical terms in this specification refer to the customary terms in the technical field. If some terms are explained or defined in this specification, the interpretation of these terms shall be based on the explanation or definition in this specification. Each embodiment of the present invention has one or more technical features. Under the premise of possible implementation, ordinary technicians in this technical field can selectively implement some or all of the technical features in any embodiment, or selectively combine some or all of the technical features in these embodiments.

[0026] In the various embodiments listed below, the same reference numerals will be used to represent the same or similar elements or components.

[0027] Figure 1 is a schematic diagram of a drug screening system according to an embodiment of the present invention. Figure 1 The drug screening system 100 includes an encoding module 110 , a candidate drug generating module 120 and a drug ranking module 130 .

[0028] The encoding module 110 encodes the drug expression and at least one drug ranking index to generate a first encoding variable 200. In this embodiment, the encoding module 110 may be, for example, an encoder. The first encoding variable is, for example, a latent variable and includes the drug expression and the drug ranking index, such as Figure 2 As shown. Figure 2 In the present invention, the drug expression is encoded in region 210, and the drug ranking index is encoded in region 220 following region 210. The drug expression and the drug ranking index can be obtained from known compounds, for example, through active assays.

[0029] In one embodiment, the drug expression includes, for example, a simplified molecular input line entry specification (SMILES) or a compound fingerprint, but the present invention is not limited thereto. The simplified molecular input line entry specification is a specification that explicitly describes the molecular structure using an ASCII string. That is, the simplified molecular linear input specification is represented as a string, such as c1c(O)c1(NC(=O)). The compound fingerprint extracts important structural features from the simplified molecular linear input specification. That is, the compound fingerprint is represented as a vector, such as [1, 1, 0, 1, 0, 0..........]. The drug expression uniquely corresponds to a particular compound, such as a real compound of MCF-7 (Breast Cancer) or other compounds.

[0030] In addition, the drug ranking index includes drug structure distribution or pharmacological properties. In one embodiment, the drug structure distribution may include, for example, the number of the nearest neighbors of the compound. For example, Figure 3 As shown, the similarity between each compound and its neighboring node is calculated. Next, based on compound 310, the number of compounds N falling within the similarity threshold K is calculated. The number of compounds N is binarized. If N>0, the value of N is set to 1. Figure 3 As shown, there are 6 compounds falling within the similarity threshold value K, so the value of N is 1. Then, the value of N is used as the drug structure distribution. In another embodiment, the pharmacological properties include, for example, IC50, GI, etc., but the present invention is not limited thereto.

[0031] The candidate drug generation module 120 is coupled to the encoding module 110. In the present embodiment, the candidate drug generation module 120 may be, for example, a microprocessor or a microcontroller, but the present invention is not limited thereto. The candidate drug generation module 120 trains a generative adversarial network (GAN) based on a first encoding variable to generate a plurality of candidate drugs, each of which has a generated drug expression and at least one generated drug ranking index. In the present embodiment, the candidate drug may be, for example, a lead compound. The generated drug expression may include, for example, a simplified molecular linear input specification (SMILES) or a compound fingerprint, but the present invention is not limited thereto. The generated drug ranking index includes a drug structure distribution or a pharmacological property, respectively. The description of the generated drug expression and the generated drug ranking index may refer to the description of the above drug expression and the drug ranking index, so it will not be repeated here.

[0032] After the candidate drug generation module 120 receives the first encoding variable generated by the encoding module 110, the candidate drug generation module 120 trains the generative adversarial network based on the first encoding variable. For example, the candidate drug generation module 120 can generate multiple fake drug expressions and multiple fake drug ranking indicators based on multiple random values. In the present embodiment, the fabricated drug expressions and their fabricated drug ranking indicators are different from each other. The fabricated drug expressions, for example, respectively include simplified molecular linear input specifications (SMILES) or compound fingerprints, but the present invention is not limited to this. The fabricated drug ranking indicators respectively include drug structure distribution or pharmacological properties. The description of the fabricated drug expression and the fabricated drug ranking indicator can refer to the description of the drug expression and the drug ranking indicator above, so it will not be repeated here.

[0033] Next, the candidate drug generation module 120 transmits the fabricated drug expression and the fabricated drug ranking index to the encoding module 110. Then, the encoding module 110 encodes the fabricated drug expression and the encoded drug ranking index to generate a plurality of second encoded variables. Then, the encoding module 110 transmits the second encoded variable to the candidate drug generation module 120. Then, the candidate drug generation module 120 can compare the first encoded variable with the second encoded variable to generate a plurality of judgment values. The judgment values ​​are used to indicate the degree of similarity between the drug expression and the drug ranking index included in the first encoded variable and the fabricated drug expression and the fabricated drug ranking index included in the second encoded variable.

[0034] Next, the candidate drug generation module 120 can determine whether the above judgment value reaches the default value. In the present embodiment, the above judgment value is, for example, 50%, but the present invention is not limited thereto. When it is determined that the above judgment value reaches the default value, it indicates that the judgment is converged, and the fabricated drug expression and the coded drug ranking index included in the second coding variable are similar to the drug expression and the drug ranking index included in the first coding variable. Next, the candidate drug generation module 120 can generate candidate drugs based on the fabricated drug expression and the fabricated drug ranking index corresponding to the second coding variable, wherein the generated drug expression corresponds to the fabricated drug expression, and the generated drug ranking index corresponds to the fabricated drug ranking index.

[0035] When it is determined that the above judgment value does not reach the default value, it means that the judgment is not convergent, and the fabricated drug expression and the fabricated drug ranking index included in the second coding variable are not similar to the drug expression and the drug ranking index included in the first coding variable. Then, the candidate drug generation module 120 can update the fabricated drug expression and the fabricated drug ranking index. For example, the candidate drug generation module 120 can modify the fabricated drug expression and the fabricated drug ranking index according to the above judgment value, so as to update the fabricated drug expression and the fabricated drug ranking index.

[0036] Then, the candidate drug generation module 120 may again transmit the updated fabricated drug expression and the updated fabricated drug ranking index to the encoding module 110. Thereafter, the encoding module 110 encodes the updated fabricated drug expression and the updated fabricated drug ranking index to generate an updated second encoding variable. Then, the candidate drug generation module 120 may compare the first encoding variable with the updated second encoding variable to generate a judgment value. Thereafter, the candidate drug generation module 120 may again determine whether the judgment value reaches a default value. Then, the candidate drug generation module 120 may repeatedly perform the above method until the judgment value reaches a default value and the candidate drug generation module 120 generates a candidate drug according to the fabricated drug expression and the fabricated drug ranking index corresponding to the second encoding variable.

[0037] The drug ranking module 130 is coupled to the candidate drug generation module 120. In the present embodiment, the drug ranking module 130 may be, for example, a microprocessor or a microcontroller, but the present invention is not limited thereto. The drug ranking module 130 sorts the strength of the candidate drugs according to the generated drug ranking index of each candidate drug to screen the drugs. That is, when the drug ranking module 130 receives the candidate drug, the drug ranking module 130 may obtain the generated drug expression and the generated drug ranking index from the candidate drug.

[0038] Next, the drug ranking module 130 can analyze the generated drug expressions and generated drug ranking indicators of the candidate drugs. For example, the drug ranking module 130 can analyze the strength of the candidate drugs based on the generated drug ranking indicators. Next, the drug ranking module 130 will sort the strength of the candidate drugs based on the strength of the generated drug ranking indicators to screen the drugs, and generate a ranking form for the candidate drugs. Therefore, the embodiment of the present invention can effectively reduce the drug development schedule and expenses, and speed up the drug development process.

[0039] In addition, the drug screening system 100 further includes a display module 140. In the present embodiment, the display module 140 may be, for example, a display or a monitor, but the present invention is not limited thereto. The display module 140 couples the coding module 110, the candidate drug generation module 120, and the drug ranking module 130. The display module 140 displays the first coding variable, the candidate drug, the generated drug expression, the generated drug ranking index, and the strength of the candidate drug through a graphical user interface. Therefore, the user can quickly learn about the operation of the drug screening system 100 and the generation of candidate drugs, thereby effectively reducing the time and cost of drug development and speeding up the development process.

[0040] Figure 4 for Figure 1 Detailed schematic diagram of the candidate drug generation module. Please refer to Figure 4 In this embodiment, the candidate drug generation module 120 includes a fabricated drug generation module 410 and a determination module 420 .

[0041] The fabricated drug generation module 410 is coupled to the encoding module 110. The fabricated drug generation module 410 generates multiple fabricated drug expressions and fabricated drug ranking indicators based on multiple random values. For example, the fabricated drug generation module 410 can generate a fabricated drug expression and a fabricated drug ranking indicator through a random value. That is to say, one random value will correspond to one fabricated drug expression and a fabricated drug ranking indicator. The fabricated drug generation module 410 can generate two fabricated drug expressions and fabricated drug ranking indicators through two random values. That is to say, two random values ​​will correspond to two fabricated drug expressions and fabricated drug ranking indicators. The fabricated drug generation module 410 can generate three fabricated drug expressions and fabricated drug ranking indicators through three random values. That is to say, three random values ​​will correspond to three fabricated drug expressions and fabricated drug ranking indicators. The rest is analogous. In addition, the fabricated drug expression and the fabricated drug ranking indicator are used to train the generative adversarial network. In this embodiment, the above-mentioned random values ​​can be preset by the user, for example. Next, the fabricated drug generation module 410 transmits the fabricated drug expression and the fabricated drug ranking index to the encoding module 110. Thereafter, the encoding module 110 may encode the fabricated drug expression and the fabricated drug ranking index to generate a second encoded variable.

[0042] The determination module 420 is coupled to the coding module 110 and the fabricated drug generation module 410. The determination module 420 receives the first coding variable and the second coding variable, and compares the first coding variable with the second coding variable to generate a determination value.

[0043] After the fabricated drug generation module 410 receives the above judgment value, the fabricated drug generation module 410 can determine whether the above judgment value reaches the default value. When it is determined that the above judgment value reaches the default value, it means that the fabricated drug expression and the fabricated drug ranking index included in the second coding variable are similar to the drug expression and the drug ranking index included in the first coding variable. Then, the fabricated drug generation module 410 can generate candidate drugs based on the fabricated drug expression and the fabricated drug ranking index corresponding to the second coding variable, wherein the generated drug expression corresponds to the fabricated drug expression, and the generated drug ranking index corresponds to the fabricated drug ranking index.

[0044] When it is determined that the above judgment value does not reach the default value, it means that the fabricated drug expression and the fabricated drug ranking index included in the second coding variable are not similar to the drug expression and the fabricated drug ranking index included in the first coding variable. Then, the fabricated drug generation module 410 can update the fabricated drug expression and the fabricated drug ranking index. For example, the fabricated drug generation module 410 can modify the fabricated drug expression and the fabricated drug ranking index according to the above judgment value, so as to update the fabricated drug expression and the fabricated drug ranking index.

[0045] Then, the fabricated drug generation module 410 may transmit the updated fabricated drug expression and the updated fabricated drug ranking index to the encoding module 110 again. After that, the encoding module 110 updates the updated fabricated drug expression and the updated fabricated drug ranking index to generate an updated second encoding variable. Then, the judgment module 420 may compare the first encoding variable with the updated second encoding variable again to generate a judgment value. The judgment value may be transmitted to the fabricated drug generation module 410. After that, the fabricated drug generation module 410 may again judge whether the judgment value reaches the default value. Then, the fabricated drug generation module 410 and the judgment module 420 may repeat the above method until the judgment value reaches the default value and the fabricated drug generation module 410 generates a candidate drug according to the fabricated drug expression and the fabricated drug ranking index.

[0046] In addition, the fabricated drug generation module 410 and the judgment module 420 are coupled to the display module 140. Accordingly, the display module 140 can further display the fabricated drug expression, the fabricated drug ranking index, the second coding variable, the above judgment value and the comparison result through a graphical user interface. Therefore, the user can quickly know the operation of the drug screening system and the generation of candidate drugs, thereby effectively reducing the time and cost of drug development and speeding up the development process.

[0047] Through the description of the above embodiments, the embodiments of the present invention provide a drug screening method. Figure 5 FIG. 4 is a flow chart of a drug screening method according to an embodiment of the present invention.

[0048] In step S502, the drug expression and at least one drug ranking index are encoded to generate a first encoding variable. In step S504, a generative adversarial network is trained based on the first encoding variable to generate a plurality of candidate drugs, wherein each candidate drug has a generated drug expression and at least one generated drug ranking index.

[0049] In step S506, the strength of the candidate drugs is ranked to screen the drugs based on at least one generated drug ranking index of each candidate drug. In step S508, the first coded variable, the candidate drug, the generated drug expression, the generated drug ranking index and the strength of the candidate drug are displayed. In this embodiment, the drug expression and the generated drug expression respectively include a simplified molecular linear input specification (SMILES) or a compound fingerprint, but the present invention is not limited to this. In addition, the drug ranking index and the generated drug ranking index respectively include drug structure distribution or pharmacological properties. The drug structure distribution includes the number of nearest neighbors of the compound, and the pharmacological properties include IC50, GI, but the present invention is not limited to this. In addition, the drug expression uniquely corresponds to a specific compound (particular compound).

[0050] Figure 6 for Figure 5 In step S602, multiple fabricated drug expressions and multiple fabricated drug ranking indicators are generated according to multiple random values. In step S604, the fabricated drug expressions and the fabricated drug ranking indicators are encoded to generate multiple second encoded variables.

[0051] In step S606 , a first coding variable and a second coding variable are received, and the first coding variable and the second coding variable are compared to generate a plurality of determination values.

[0052] In step S608, it is determined whether the judgment value reaches the default value. When it is determined that the judgment value reaches the default value, step S610 is entered to generate candidate drugs according to the fabricated drug expression corresponding to the second coding variable and the fabricated drug ranking index.

[0053] When it is determined that the judgment value does not reach the default value, step 612 is entered to update the fabricated drug expression and the fabricated drug ranking index. Then, after executing step S612, the above method enters step S606, and the drug expression and the drug ranking index are compared again with the updated fabricated drug expression and the updated fabricated drug ranking index to generate a judgment value. In this embodiment, the fabricated drug expression includes a simplified molecular linear input specification (SMILES) or a compound fingerprint, but the present invention is not limited to this. In addition, the fabricated drug ranking index includes drug structure distribution or pharmacological properties. The drug structure distribution includes the number of nearest neighbors of the compound, and the pharmacological properties include IC50 and GI, but the present invention is not limited to this.

[0054] In summary, the drug screening system and drug screening method disclosed in the present invention encode drug expressions and drug ranking indicators to generate a first encoding variable. Based on the first encoding variable, a generative adversarial network is trained to generate multiple candidate drugs, each of which has a generated drug expression and at least one generated drug ranking indicator. Based on the generated drug ranking indicator of each candidate drug, the strength of the candidate drugs is ranked to screen the drugs. In this way, the embodiments of the present invention can effectively reduce the time and cost of drug development and speed up the development process.

[0055] Although the present invention is disclosed as above by the embodiments, it is not intended to limit the scope of the present invention. Any ordinary technician in the technical field can make some changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be based on that defined in the claims.

Claims

1. A drug screening system, comprising: An encoding module encodes a drug expression and at least one drug ranking index to generate a first encoding variable, wherein the first encoding variable is a latent variable and includes the drug expression and the at least one drug ranking index; A candidate drug generation module, which trains a generative adversarial network according to the first encoding variable to generate a plurality of candidate drugs, wherein each candidate drug has a generated drug expression and at least one generated drug ranking index; as well as A drug ranking module, which ranks the strengths of the candidate drugs to screen the drugs according to the at least one generated drug ranking index of each candidate drug; Among them, the candidate drug generation module includes: a fabricated drug generation module, which generates a plurality of fabricated drug expressions and a plurality of fabricated drug ranking indicators to the encoding module according to a plurality of random values; and a determination module, receiving the first coding variable and a plurality of second coding variables, wherein the second coding variables are generated by the coding module according to the fabricated drug expressions and the fabricated drug ranking indicators, and the determination module compares the first coding variable with the second coding variables to generate a plurality of determination values; Among them, the fabricated drug generation module determines whether these judgment values ​​reach a default value. When it is determined that these judgment values ​​reach the default value, the fabricated drug generation module generates these candidate drugs based on these fabricated drug expressions corresponding to the second coding variable and these fabricated drug ranking indicators. When it is determined that these judgment values ​​do not reach the default value, the fabricated drug generation module updates these fabricated drug expressions and these fabricated drug ranking indicators.

2. The drug screening system as described in claim 1, wherein the drug expression comprises a simplified molecular linear input specification or a compound fingerprint, and the generated drug expression comprises the simplified molecular linear input specification or the compound fingerprint. 3 . The drug screening system as claimed in claim 1 , wherein the at least one drug ranking index and the at least one generated drug ranking index respectively comprise a drug structure distribution or a pharmacological property. 4 . The drug screening system as claimed in claim 3 , wherein the drug structure distribution includes the number of nearest neighbors of a compound, and the pharmacological properties include IC50 and GI.

5. The drug screening system as claimed in claim 1, wherein the fabricated drug expressions respectively include a simplified molecular linear input specification or a compound fingerprint. 6 . The drug screening system as claimed in claim 1 , wherein the fabricated drug ranking indicators respectively include a drug structure distribution or a pharmacological property.

7. The drug screening system as claimed in claim 6, wherein the drug structure distribution includes the number of nearest neighbors of a compound, and the pharmacological properties include IC50 and GI.

8. The drug screening system as claimed in claim 1, wherein the drug expression uniquely corresponds to a specific compound.

9. The drug screening system as claimed in claim 1, wherein the drug screening system further comprises: A display module is coupled to the encoding module, the candidate drug generating module and the drug ranking module, and the display module displays the first encoding variable, the candidate drugs, the generated drug expressions, the generated drug ranking indicators and the strengths of the candidate drugs.

10. A drug screening method comprising: Encoding a drug expression and at least one drug ranking index to generate a first encoding variable, wherein the first encoding variable is a latent variable and includes the drug expression and the at least one drug ranking index; According to the first encoding variable, a generative adversarial network is trained to generate a plurality of candidate drugs, wherein each candidate drug has a generated drug expression and at least one generated drug ranking index; as well as According to the at least one generated drug ranking index of each candidate drug, the strength of the candidate drugs is ranked to screen the drugs; The step of training the generative adversarial network to generate the candidate drugs according to the first encoding variable comprises: According to the multiple random values, multiple fabricated drug expressions and multiple fabricated drug ranking indicators are generated; encoding the fabricated drug expressions and the fabricated drug ranking indicators to generate a plurality of second coded variables; receiving the first coding variable and the second coding variables, and comparing the first coding variable with the second coding variables to generate a plurality of determination values; Determine whether these judgment values ​​reach a default value; When it is determined that the judgment values ​​reach the default value, generating the candidate drugs according to the fabricated drug expressions corresponding to the second coding variable and the fabricated drug ranking indicators; and When it is determined that the judgment values ​​do not reach the default values, the fabricated drug expressions and the fabricated drug ranking indices are updated.

11. The drug screening method as claimed in claim 10, wherein the drug expression comprises a simplified molecular linear input specification or a compound fingerprint, and the generated drug expression comprises the simplified molecular linear input specification or the compound fingerprint. 12 . The drug screening method as claimed in claim 10 , wherein the at least one drug ranking index and the at least one generated drug ranking index respectively comprise a drug structural distribution or a pharmacological property.

13. The drug screening method as claimed in claim 12, wherein the drug structure distribution includes the number of nearest neighbors of a compound, and the pharmacological properties include IC50 and GI.

14. The drug screening method as claimed in claim 10, wherein the fabricated drug expressions respectively comprise a simplified molecular linear input specification or a compound fingerprint.

15. The drug screening method as claimed in claim 10, wherein the fabricated drug ranking indicators respectively include a drug structure distribution or a pharmacological property.

16. The drug screening method as claimed in claim 15, wherein the drug structure distribution includes the number of nearest neighbors of a compound, and the pharmacological properties include IC50 and GI.

17. The drug screening method as claimed in claim 10, wherein the drug expression uniquely corresponds to a specific compound.

18. The drug screening method according to claim 10, wherein the drug screening method further comprises: The first coded variable, the candidate drugs, the generated drug expressions, the generated drug ranking indices and the strengths of the candidate drugs are displayed.

Citation Information

Patent Citations

  • Lead compound virtual screening method and device

    CN107862173A

  • Method and system for screening disease drug targets and target combinations

    CN108647489A

  • A method and a system for virtual design and evaluation of pharmaceutical formulation

    CN108984811A

  • Generative machine learning systems for drug design

    US20170161635A1

  • Systems and methods for multimodal generative machine learning

    US20190018933A1