Available online on 15.07.2025 at http://jddtonline.info

Journal of Drug Delivery and Therapeutics

Open Access to Pharmaceutical and Medical Research

Copyright  © 2025 The  Author(s): This is an open-access article distributed under the terms of the CC BY-NC 4.0 which permits unrestricted use, distribution, and reproduction in any medium for non-commercial use provided the original author and source are credited

Open Access Full Text Article                                                              Review Article

The role of Artificial Intelligence in drug discovery

Samiksha B. Rotake*, Pooja R. Hatwar , Dr. Ravindra L. Bakal , Shruti I. Meshram

Shri Swami Samarth Institute of Pharmacy, At Parsodi, Dhamangaon Rly (444709), Dist. Amravati, Maharashtra, India

 

 

Introduction: 

In recent years, several ideas related to artificial intelligence (AI) have been successfully applied in computer-assisted drug discovery ¹. The influence of artificial intelligence technologies on societal life has notably surged within the last decade ². Artificial intelligence (AI) showcases adaptability throughout numerous phases of the drug discovery process ³. Both AI and machine learning technologies are instrumental in shaping this field ⁴.

Drug discovery has embraced emerging technologies like Utilizing data science, informatics, and AI can expedite the development of effective treatments, leading to reduced costs and fewer animal experiments. This shift is observable through rising enthusiasm from investors, industry experts, academic researchers, and policymakers ⁵. Successful drug discovery process, and technologies related to machine learning and artificial intelligence are crucial in this area ⁶. Technology’s significance spans across all facets of life, including drug discovery endeavors ⁷. The primary stage of drug discovery and development involves target identification and AI’s expanding impact extends across numerous areas linked to drug discovery ⁸. 

However, for widespread adoption by the medicinal chemistry community, it remains critical to develop AI models achieving high predictive performance while being explainable to end-users based on their knowledge and background ¹⁰. AI operates using systems and software capable of learning from data to independently make informed judgments in pursuit of specific goals ¹¹. Since the 1960s, artificial intelligence (AI) has been used continuously in drug development at several stages, including target identification, lead optimization, and drug design ^12,13. Recent years have witnessed a rapid surge in using AI algorithms and big data in drug discovery, resulting in reduced human workload and considerable improvements in quality of life ^14,15,16.

Advancements in AI technology significantly contribute to initial market analysis reports and accelerate research work in drug discovery ^17,18. The union of AI within the intricate landscape of drug discovery embodies an unprecedented convergence of technology and biology ¹⁹. The process of discovering new drugs is time-consuming and costly, necessitating several steps to ensure the quality, safety, and potency of the marketed drug ²⁰. Potency against biological targets, selectivity against undesirable targets, and advantageous physicochemical and ADMET qualities are just a few of the requirements that clinical candidate compounds must meet in the drug discovery process ²¹. Consequently, compound optimization remains a multidimensional challenge. The integration of artificial intelligence (AI) with novel experimental technologies is anticipated to expedite, reduce costs, and enhance the efficacy of the search for new pharmaceuticals ^22,23. AI, in its diverse forms, is currently successfully applied in numerous domains and complex tasks, spanning robotics, speech translation, image analysis, logistics, and ongoing efforts in molecule design ²⁴.

Drug Discovery involves a sequence encompassing candidate identification, synthesis, characterization, screening, and efficacy assays. Following successful testing, drug development ensues before clinical trials. This stage holds particular importance for the future, ensuring robust binding between the drug molecule and its target proteins or enzymes. Conventionally, discovering new drugs takes about 10-12 years, encompassing various research and development stages. The timeline flow diagram below illustrates the traditional drug discovery method. Each stage requires over a year, culminating in an overall drug approval timeframe of approximately 10-15 years ^{25, 26}.

This study highlights the achievements in artificial intelligence, emphasizing its role in improving prior approaches to drug discovery ²⁷.

 

 

 

Figure 1: Traditional drug discovery timeline ²⁸

Artificial intelligence (AI):

Artificial intelligence (AI) is one of our era's swiftly advancing technologies. It represents a domain within computer science dedicated to creating intelligent systems ²⁹.

Classification of AI: ³⁰

Figure 2: Types of Artificial Intelligence ³⁰

 

Artificial intelligence in drug discovery:

Artificial Intelligence (AI) has significantly impacted drug discovery ³¹, particularly in the challenging phase of finding new drugs, which has historically been the most arduous aspect since the outset of the drug discovery process ³². The drug discovery and development process holds significant importance in identifying fresh therapeutic targets, screening potential lead compounds, and assessing drug efficacy and safety ³³. In the quest for anti-cancer drugs, the screening process often consumes considerable time. To expedite this process, Novartis utilized machine learning algorithms and images to predict potential subjects worth further exploration, capitalizing on the swift analytical capabilities of computers compared to traditional human analysis and laboratory tests ³⁴. This not only accelerates the discovery of promising compounds but also mitigates the labor costs linked to manually verifying each compound. 

Cutting-edge intellectual property like the Jal Action Platform within leading biopharmaceutical companies holds the promise of enhancing health outcomes by enabling real-time data collection and integration in the quest for discoveries, often incorporating Wetware technology ³⁵. Also, after identifying potential targets drug ability, discovering drugs that interact with these therapeutic targets is crucial ³⁶.

Given the complexity of treating concomitant diseases that might necessitate multiple drugs, predicting interactions between drug-target and drug-drug becomes essential to minimize the risks of increased side effects ³⁷. A transformative era in drug discovery research has been initiated by the development of AI, which has advanced from conventional trial-and-error or hypothesis-driven methods to more rational, data-driven approaches ³⁸. AI and data-driven approaches, including AI are reshaping drug discovery processes ³⁹. 

Recent times have witnessed a renewed interest in incorporating deep learning into drug discovery, resulting in a remarkable upsurge in innovative modeling approaches and applications ⁴⁰. In drug discovery, AI techniques primarily fall into two categories: supervised learning and unsupervised learning. Unsupervised learning methods excel in exploratory data analysis by revealing concealed patterns within unlabeled data or facilitating data clustering ⁴¹. The diverse spectrum of AI applications in drug discovery is depicted in the accompanying diagram. Despite its inherent advantages, AI grapples with significant data-related challenges such as managing issues related to data scale, expansion, diversity, and uncertainty ⁴².

 

 

Figure 3: Drug discovery cycle ⁴³.

Figure 4: Role of artificial intelligence (AI) in drug discovery ⁴⁴.

 

The process of drug discovery involves multiple intricate stages, generally categorized into four key phases: (i) target selection and validation; (ii) compound screening and lead optimization; (iii) preclinical studies; and (iv) clinical trials. Initially, identifying the target associated with a particular disease necessitates various evaluations, including cellular and genetic target assessments, genomic and proteomic analyses, as well as bioinformatic predictions ⁴⁵. The subsequent stage involves hit identification, where compounds are sourced from molecular libraries via techniques like combinatorial chemistry, high-throughput screening, and virtual screening. This phase employs a cyclic process of structure-activity studies, in silico analyses, and cellular functional tests to enhance the functional attributes of newly synthesized drug candidates ⁴⁶.

Following this, in vivo studies, including pharmacokinetic assessments and toxicity tests, are conducted using animal models. Finally, the drug candidate, having passed through preclinical assessments, enters clinical trials involving three sequential phases: Phase I for testing drug safety with a small human cohort, Phase II for evaluating drug efficacy with a limited number of individuals affected by the targeted disease, and Phase III for conducting efficacy studies with a larger patient group. Approval and commercialization of the compound by regulatory bodies like the FDA follow if its safety and efficacy are confirmed during these clinical phases. Notably, the traditional drug discovery pipeline incurs an average cost of around 2.6 billion USD, with a timeline extending over 12 years ⁴⁷.

Drug Discovery and Development Process: ^48,49

The drug discovery and development process include distinct stages: 

1. Target selection and validation, a crucial phase involving the identification of disease-specific targets.
2. Compound screening and lead optimization, focused on refining drug leads sourced from molecular libraries to improve their desired properties.
3. Pre-clinical studies, comprising pharmacokinetic assessments and toxicity tests performed in animal models.
4.  Clinical trials, where the potential drug candidate is administered to volunteers to assess any adverse effects.
5. Final approval of the drug for commercialization.

Advantages of AI In drug discovery: ^51,52

1.  AI plays a crucial role in predicting environmental factors that can impact drug stability and shelf life, leading to improved formulation and durability of pharmaceuticals.
2. The implementation of AI significantly reduces the time and cost involved in medication development, making the entire research and development process more cost-effective.
3. AI contributes to error reduction and enhanced accuracy, utilizing intelligent robots for space exploration due to their robust construction and ability to withstand harsh conditions.
4. Industries such as mining and fuel exploration benefit from AI technology by rectifying human errors and enabling exploration in challenging environments, including oceanic studies.
5. AI facilitates everyday tasks; for instance, GPS aids in long commutes, while AI-powered androids predict user inputs and assist with tasks like spelling correction.
6. Modern businesses employ advanced AI systems, such as ‘avatar’ digital assistants, to reduce human intervention. These systems rely on logic and reason, avoiding the influence of human emotions, thereby enhancing judgment.
7. Medical applications of AI enable doctors to evaluate medication side effects and health risks, assisting in patient assessment and healthcare decisions without direct human intervention.

Disadvantages of AI in drug discovery: ⁵²

1. AI-equipped robots possess the capacity to mimic human thought processes while remaining detached from emotions, enabling more precise task execution without subjective judgment. However, AI lacks the capability to enhance human resources, as it cannot discern between diligent and inactive individuals based on experience.
2. Unlike AI technology, humans have the innate ability to exercise creativity and original thought processes that are beyond the current capabilities of artificial intelligence.
3. The widespread implementation of AI technology may lead to substantial unemployment, altering established work patterns and potentially stifling human workers creativity and innovation.
4. Implementing AI machines involves intricate design, maintenance, and repair processes that demand significant time and resources from the R&D sector. Regular software updates, machine reinstalls, and recovery procedures also contribute to the substantial time and financial investment in AI technology.

Application:

AI technology has found extensive application across the entirety of drug Research and Development (R&D), encompassing target identification, drug design, synthesis, and property evaluation. Undoubtedly, it has significantly reduced the drug R&D timeline and substantially reduced experimental costs when compared to conventional experimental procedures. Ongoing efforts by scientists to further integrate AI into various domains aim to foster the advancement of pharmaceutical sciences. AI’s value in the drug discovery market is projected to grow significantly, with a compound annual growth rate of 40.8%, surging from 260 million USD globally in 2019 to an estimated 1.43 billion USD in 2024 ⁵³.

In drug discovery, target identification aims at pinpointing molecules, commonly proteins, capable of altering disease states through modulation of their activity ⁵⁴. Machine learning algorithms possess the capability to analyze diverse datasets, including gene expression profiles, protein-protein interaction networks, and genomic and proteomic data. These algorithms aid in the identification of potential targets believed to play roles in disease pathways. Despite the vast human proteome comprising around 20,000 proteins, only about 3,000 have been recognized as potential therapeutic targets. Future advancements in knowledge might further augment our comprehension of proteins that could potentially serve as drug targets ⁵⁵.

The pharmaceutical development research sector has witnessed a rising demand for in silico technologies and AI applications. This trend is expected to persist due to the necessity for precise forecasts of pharmacokinetics/ADMET (Absorption, Distribution, Metabolism, and Toxicity) concerning hit compounds. Evaluation of pharmacokinetics and toxicity remains crucial to prevent the potential failure of candidate drugs in later stages of drug development ⁵⁶.

During this phase, a blend of favorable characteristics was pinpointed, prompting the initiation of clinical trials. Human volunteers serve as test subjects at this stage ⁵⁷.

Natural Language Processing (NLP) finds applications in mining scientific literature to extract information regarding adverse effects, including drug toxicity or resistance, facilitating automated assessments essential for regulatory (FDA) approval or patent applications ⁵⁸. Additionally, NLP-based sentiment analysis techniques can suggest drugs ⁵⁹. Furthermore, machine learning-based systems capable of predicting potential product sales assist pharmaceutical companies in optimizing their business resources. The FDA acknowledges the emerging issues with 3D printing and is actively researching to comprehensively understand the technology ⁶⁰.

5. Preclinical studies, focusing on pharmacology, pharmacokinetics, and toxicology, rely on understanding a drug’s physicochemical properties and ADMET characteristics ⁶¹.
6. Pediatric patients unique needs for safe drug products and adherence require specific considerations in drug design due to limited clinical trials for pediatric applications ⁶².
7. Compound solid preparations, composed of multiple parts with varying amounts of medicine, undergo processes like pulverization, mixing, screening, drying, and molding to form solid medicine preparations ⁶³.
8. AI assists in developing complex formulations, predicting physicochemical parameters, and optimizing drug stability, solubility, and viscosity for injectable and biologics, enhancing formulation development ⁶⁴.

Challenges and limitations of AI in drug discovery:

The term “Artificial intelligence in drug discovery” may imply successful implementation of AI in early drug discovery stages ⁶⁵. However, while AI-driven drug discovery holds significant promise, it faces several persistent challenges and limitations. The interpretability of deep learning models poses a concern as comprehending the reasoning behind model predictions is vital to establish trust in these predictions. Furthermore, issues related to data quality, standardization, ethical implications, and regulatory acceptance of AI-driven drug discovery methods are crucial aspects that demand attention ⁶⁶.

Despite the existing potential benefits of AI in drug discovery, various challenges and limitations need careful consideration ⁶⁷. A primary challenge involves the availability of appropriate data ⁶⁸. AI-driven approaches typically necessitate a vast amount of data for effective training ⁶⁹.

Inefficient data integration presents another issue arising from the divergence between datasets containing raw data, processed data, metadata, or candidate data ⁷⁰.

Successful applications of AI in drug discovery:

Over the past few years, multiple reviews have highlighted the increasing importance of AI in drug discovery. Therefore, this specific review intends to focus on a range of noteworthy studies that demonstrate the practical influence of AI in small-molecule drug discovery ⁷¹.

Conflict of Interest: The authors declare no potential conflict of interest with respect to the contents, authorship, and/or publication of this article.

Author Contributions: All authors have equal contribution in the preparation of manuscript and compilation.

Source of Support: Nil

Funding: The authors declared that this study has received no financial support.

Informed Consent Statement: Not applicable. 

Data Availability Statement: The data supporting in this paper are available in the cited references. 

Ethical approval: Not applicable.

References:

1. Jiménez-Luna J, Grisoni F, Schneider G. Drug discovery with explainable artificial intelligence. Nat Mach Intell. 2020; 2(11):573–584. DOI: http://dx.doi.org/10.1038/s42256-020-00236-4
2. Karger E, Kureljusic M. Using artificial intelligence for drug discovery: A bibliometric study and future research agenda. Pharmaceuticals. 2022; 15(12):1492. DOI: http://dx.doi.org/10.3390/ph15121492
3. Gupta R, Srivastava D, Sahu M, et al. Artificial intelligence to deep learning: machine intelligence approach for drug discovery. Mol Divers. 2021; 25:1315–1360. DOI: http://dx.doi.org/10.1007/s11030-021-10217-3
4. Mbatha SZ, et al. Artificial intelligence–enhanced drug discovery and the achievement of next-generation human-centered health system. In: Dlamini Z, editor. Society 5.0 and next generation healthcare, Cham: Springer; 2023. DOI: http://dx.doi.org/10.1007/978-3-031-36461-7_7
5. Vidhya KS, Sultana A, M N, et al., Artificial intelligence’s impact on drug discovery and development from bench to bedside. Cureus. 2023; 15(10):e47486. DOI: http://dx.doi.org/10.7759/cureus.47486
6. Hasselgren C, Oprea TI. Artificial intelligence for drug discovery: are we there yet? Annu Rev Pharmacol Toxicol. 2024; 64. DOI: http://dx.doi.org/10.1146/annurev-pharmtox-040323-040828
7. Mathew A, Alex H. From code to cure: the role of AI in accelerating drug discovery. Advances and Challenges in Science and Technology, 2023; 2:94–102. DOI: http://dx.doi.org/10.9734/bpi/acst/v2/19866D
8. Sharma V, Singh A, Chauhan S, et al. Role of artificial intelligence in drug discovery and target identification in cancer. Curr Drug Deliv. 2023; 20(10). DOI: http://dx.doi.org/10.2174/1567201821666230905090621
9. Sahayasheela VJ, Lankadasari MB, Dan VM, Dastager SG, Pandian GN. Artificial intelligence in microbial natural product drug discovery: current and emerging role. Nat Prod Rep. 2022; 39(12):2215–2230.
10. Ponzoni I, Páez Prosper JA, Campillo NE. Explainable artificial intelligence: a taxonomy and guidelines for its application to drug discovery. WIREs Comput Mol Sci. 2023; 13(6):e1681. DOI: http://dx.doi.org/10.1002/wcms.1681
11. Habeeba S. Use of artificial intelligence in drug discovery and its application in drug development Asian J Res Chem. 2023; 16(1). DOI: http://dx.doi.org/10.52711/0974-4150.2023.00014
12. Mahjoub MA, Sheikholislam Z. Artificial intelligence in drug discovery and delivery: advancements and applications. J Basic Res Eng Sci. 2023; 4(7):1140–1142. DOI: http://dx.doi.org/10.37871/jbres1778
13. Mhatre G. Artificial intelligence in drugs discovery and development. Int J Sci Res Eng Manag. 2023; 7(4). DOI: http://dx.doi.org/10.55041/IJSREM18863
14. Nguyen MH, Tran ND, Le NQK. Big data and artificial intelligence in drug discovery for gastric cancer: current applications and future perspectives. Curr Med Chem. 2023; 30(40). DOI: http://dx.doi.org/10.2174/0929867331666230913105829
15. Patel V, Shah M. A comprehensive study on artificial intelligence and machine learning in drug discovery and drug development. Intelligent Medicine. 2021; 2(11). DOI: http://dx.doi.org/10.1016/j.imed.2021.10.001
16. Bijral RK, Singh I, Manhas J, Sharma V. Exploring artificial intelligence in drug discovery: a comprehensive review. Arch Comput Methods Eng. 2021; 29(5):1–17. DOI: http://dx.doi.org/10.1007/s11831-021-09661-z
17. Yadav V, Reang J, Vinita, Tonk RK. Artificial intelligence and machine learning in drug discovery. In: Kumar V, editor. CADD and informatics in drug discovery. Delhi Pharmaceutical Sciences and Research University Delhi; 2023. P. 205–230. DOI: http://dx.doi.org/10.1007/978-981-99-1316-9_9
18. Terranova N, Renard D, Shahin MH, et al. Artificial intelligence for quantitative modeling in drug discovery and development: an Innovation and Quality Consortium perspective on use cases and best practices. Clin Pharmacol Ther. 2023; Early View. DOI: http://dx.doi.org/10.1002/cpt.3053
19. Ashiwaju BI, Orikpete OF, Uzougbo CG. The intersection of artificial intelligence and big data in drug discovery: a review of current trends and future implications. Matrix Sci Pharma. 2023; 7(2):36–42. DOI: http://dx.doi.org/10.4103/mtsp.mtsp_14_23
20. Barsagde N, Baranga T, Shembekar S, Kurani K, Chakraborty I, Soni A. Artificial intelligence for drug discovery. Int Res J Modern Eng Technol Sci. 2023; 5(5):1–6. DOI: http://dx.doi.org/10.56726/IRJMETS38887
21. Hessler G, Baringhaus KH. Artificial intelligence in drug design. Molecules. 2018; 23(10):2520. DOI: http://dx.doi.org/10.3390/molecules23102520
22. Chalasani SH, Syed J, Ramesh M, Patil V, Kumar TMP. Artificial intelligence in the field of pharmacy practice: a literature review. Explor Res Clin Soc Pharm. 2023; 12:100346. DOI: http://dx.doi.org/10.1016/j.rcsop.2023.100346
23. Chan HCS, Shan H, Dahoun T, Vogel H, Yuan S. Advancing drug discovery via artificial intelligence.  Trends Pharmacol Sci. 2019; 40(8):592–604. DOI: http://dx.doi.org/10.1016/j.tips.2019.06.004
24. Sellwood MA, Ahmed M, Segler MHS, Brown N. Artificial intelligence in drug discovery. Future Med Chem. 2018; 10(17):2025–2028. DOI: http://dx.doi.org/10.4155/fmc-2018-0212
25. Wang H, Fu T, Du Y, et al. Scientific discovery in the age of artificial intelligence. Nature. 2023; 620:47–60. DOI: http://dx.doi.org/10.1038/s41586-023-06221-2
26. Gautam V, Gaurav A, Masand N, et al. Artificial intelligence and machine-learning approaches in structure and ligand-based discovery of drugs affecting central nervous system. Mol Divers. 2023; 27:959–985. DOI: http://dx.doi.org/10.1007/s11030-022-10489-3
27. Madani BM. The application of artificial intelligence in antibiotic discovery: an overview of current and future perspective. Int J Adv Res. 2023; 11(4):1326–1332. DOI: http://dx.doi.org/10.21474/IJAR01/16794
28. Pakala A, Vora K, Rani BA, Dsouza OD, Balaji SM, BC S. Drug discovery using artificial intelligence. Int J All Res Educ Sci Methods (IJARESM). 2023; 11(2):1452.
29. Annammadevi GS, Pradhan M, Choudhury AK. Artificial intelligence the futuristic technology in the drug discovery process: a review. J Young Pharm. 2023; 15(3):390–396. DOI: http://dx.doi.org/10.5530/jyp.2023.15.54
30. Rao SV, Padma KS, Purna KA, Padmalatha K. Artificial intelligence and telimedicine: an effective digitalization in healthcare and pharmaceutical sectors. Eur Chem Bull. 2023; 12(Special Issue 10):3428–3443. DOI: http://dx.doi.org/10.48047/ecb/2023.12.si10.00392
31. Zhavoronkov A, Ivanenkov YA, Aliper A, et al. Deep learning enables rapid identification of potent DDR1 kinase inhibitors. Nat Biotechnol. 2019; 37(9):1038–1040. DOI: https://doi.org/10.1038/s41587-019-0224-x.
32. Walters WP, Murcko M. Assessing the impact of generative AI on medicinal chemistry. Nat Biotechnol. 2020; 38(2):143–145. DOI: https://doi.org/10.1038/s41587-019-0390-3.
33. Stokes JM, Yang K, Swanson K, et al. A deep learning approach to antibiotic discovery. Cell, 2020; 180(4):688–702.e13. DOI: https://doi.org/10.1016/j.cell.2020.01.021.
34. Yang X, Zhang J, Song Y, et al. DeepChem: a deep learning framework for drug discovery. J Chem Inf Model. 2017; 57(11):2697–2700. DOI: https://doi.org/10.1021/acs.jcim.7b00503.
35. Gao W, Coley CW. The synthesizability of molecules proposed by generative models. J Chem Inf Model, 2020; 60(12):5714–5723. DOI: https://doi.org/10.1021/acs.jcim.0c00622.
36. Walters WP, Barzilay R. Applications of deep learning in molecule generation and molecular property prediction. Chem Rev. 2020; 120(17): 12685–12727. DOI: https://doi.org/10.1021/acs.chemrev.9b00726.
37. Chen H, Engkvist O, Wang Y, Olivecrona M, Blaschke T. The rise of deep learning in drug discovery. Drug Discov Today. 2018; 23(6):1241–1250. DOI: https://doi.org/10.1016/j.drudis.2018.01.039.
38. Gawehn E, Hiss JA, Schneider G. Deep learning in drug discovery. Mol Inform. 2016; 35(1):3–14. DOI: https://doi.org/10.1002/minf.201501008.
39. Zhang Q, Ma X, Wang H, et al. A deep learning framework for compound–protein interaction prediction. Bioinformatics. 2020; 36(9):2829–2837. DOI: https://doi.org/10.1093/bioinformatics/btaa132.
40. Feinberg EN, Sur D, Wu Z, et al. PotentialNet for molecular property prediction. ACS Cent Sci. 2018; 4(11):1520–1530. DOI: https://doi.org/10.1021/acscentsci.8b00507.
41. Li X, Fourches D. Inductive transfer learning for molecular activity prediction: Next-Gen QSAR models with deep neural networks. J Cheminform. 2020; 12(1):62. DOI: https://doi.org/10.1186/s13321-020-00467-6.
42. Ekins S, Puhl AC, Zorn KM, et al. Exploiting machine learning for end-to-end drug discovery and development. Nat Mater. 2019; 18(5):435–441. DOI: https://doi.org/10.1038/s41563-019-0338-z.
43. Gawehn E, Hiss JA, Schneider G. Machine learning in chemoinformatics and drug discovery. Mol Inform. 2016; 35(1):3–14. DOI: https://doi.org/10.1002/minf.201501008.
44. Hu W, Liu B, Gomes J, Zitnik M, Liang P, Leskovec J. Strategies for pre-training graph neural networks. arXiv preprint. 2019; arXiv:1905.12265.
45. Vamathevan J, Clark D, Czodrowski P, et al. Applications of machine learning in drug discovery and development. Nat Rev Drug Discov. 2019; 18(6):463–477. DOI: https://doi.org/10.1038/s41573-019-0024-5.
46. Chenthamarakshan V, Das P, Padhi I, et al. Target-specific and selective drug design using deep generative models. Chem Sci. 2020; 11(10): 2700–2708. DOI: https://doi.org/10.1039/C9SC03754E.
47. Gaulton A, Hersey A, Nowotka M, et al. The ChEMBL database in 2017. Nucleic Acids Res. 2017; 45(D1):D945–D954. DOI: https://doi.org/10.1093/nar/gkw1074.
48. Blaschke T, Olivecrona M, Engkvist O, Bajorath J, Chen H. Application of generative autoencoder in de novo molecular design. Mol Inform. 2018; 37(1-2):1700123. DOI: https://doi.org/10.1002/minf.201700123.
49. Jaques N, Gu S, Bahdanau D, et al. Sequence tutor: conservative fine-tuning of sequence generation models with KL-control. arXiv preprint. 2017; arXiv:1611.02796.
50. Kotsias PG, Arús-Pous J, Chen H, et al. Direct steering of de novo molecular generation with descriptor conditional recurrent neural networks. Nat Mach Intell. 2020; 2(5):254–265. DOI: https://doi.org/10.1038/s42256-020-0174-3.
51. Maziarka Ł, Pocha A, Kaczmarczyk J, Rataj K, Danel T, Warchoł M. Mol-CycleGAN: a generative model for molecular optimization. J Cheminform. 2020; 12(1):2. DOI: https://doi.org/10.1186/s13321-019-0395-x.
52. Winter R, Montanari F, Noé F, Clevert D-A. Learning continuous and data-driven molecular descriptors by translating equivalent chemical representations. Chem Sci. 2019; 10(6):1692–1701. DOI: https://doi.org/10.1039/C8SC04175J.
53. Jin W, Barzilay R, Jaakkola T. Hierarchical generation of molecular graphs using structural motifs. ICML. 2020; PMLR 119:4839–4848.
54. Popova M, Isayev O, Tropsha A. Deep reinforcement learning for de novo drug design. Sci Adv. 2018; 4(7):eaap7885. DOI: https://doi.org/10.1126/sciadv.aap7885.
55. Winter R, Montanari F, Noé F, Clevert D-A. Learning continuous and data-driven molecular descriptors by translating equivalent chemical representations. Chem Sci. 2019; 10(6):1692–1701. DOI: https://doi.org/10.1039/C8SC04175J.
56. Liu Q, Allamanis M, Brockschmidt M, Gaunt A. Constrained graph variational autoencoders for molecule design. NeurIPS. 2018; 31:7795–7804.
57. Lim J, Ryu S, Kim J, Kim W-Y. Molecular generative model based on conditional variational autoencoder for de novo molecular design. J Cheminform. 2018; 10(1):31. DOI: https://doi.org/10.1186/s13321-018-0286-7.
58. Segler MH, Kogej T, Tyrchan C, Waller MP. Generating focused molecule libraries for drug discovery with recurrent neural networks. ACS Cent Sci. 2018; 4(1):120–131. DOI: https://doi.org/10.1021/acscentsci.7b00512.
59. Merk D, Friedrich L, Grisoni F, Schneider G. De Novo Design of Bioactive Small Molecules by Artificial Intelligence. Mol Inform. 2018; 37(1-2):1700153. DOI: https://doi.org/10.1002/minf.201700153.
60. Gupta A, Müller AT, Huisman BJH, Fuchs JAS, Schneider P, Schneider G. Generative recurrent networks for de novo drug design. Mol Inform. 2018; 37(1-2):1700111. DOI: https://doi.org/10.1002/minf.201700111.
61. Zhavoronkov A, Aliper A, Ivanenkov YA, et al. Deep learning enables rapid identification of potent DDR1 kinase inhibitors. Nat Biotechnol. 2019; 37(9):1038–1040. DOI: https://doi.org/10.1038/s41587-019-0224-x.
62. Gómez-Bombarelli R, Wei JN, Duvenaud D, et al. Automatic chemical design using a data-driven continuous representation of molecules. ACS Cent Sci. 2018; 4(2):268–276. DOI: https://doi.org/10.1021/acscentsci.7b00572.
63. Olivecrona M, Blaschke T, Engkvist O, Chen H. Molecular de-novo design through deep reinforcement learning. J Cheminform. 2017; 9(1):48. DOI: https://doi.org/10.1186/s13321-017-0235-x.
64. Popova M, Isayev O, Tropsha A. Deep reinforcement learning for de novo drug design. Sci Adv. 2018; 4(7):eaap7885. DOI: https://doi.org/10.1126/sciadv.aap7885.
65. Jin W, Barzilay R, Jaakkola T. Junction tree variational autoencoder for molecular graph generation. ICML. 2018; 80:2323–2332.
66. Segler MH, Preuss M, Waller MP. Planning chemical syntheses with deep neural networks and symbolic AI. Nature. 2018; 555(7698):604–610. DOI: https://doi.org/10.1038/nature25978.
67. Kotsias PG, Arús-Pous J, Chen H et al. Direct steering of de novo molecular generation with descriptor conditional recurrent neural networks. Nat Mach Intell. 2020; 2(5):254–265. DOI: https://doi.org/10.1038/s42256-020-0174-3.
68. Maziarka Ł, Pocha A, Kaczmarczyk J, Rataj K, Danel T, Warchoł M. Mol-CycleGAN: a generative model for molecular optimization. J Cheminform. 2020; 12(1):2. DOI: https://doi.org/10.1186/s13321-019-0395-x.
69. Winter R, Montanari F, Noé F, Clevert D-A. Learning continuous and data-driven molecular descriptors by translating equivalent chemical representations. Chem Sci. 2019; 10(6):1692–1701. DOI: https://doi.org/10.1039/C8SC04175J.
70. Li X, Fourches D. Inductive transfer learning for molecular activity prediction: Next-Gen QSAR models with deep neural networks. J Cheminform. 2020; 12(1):62. DOI: https://doi.org/10.1186/s13321-020-00467-6.
71. Vamathevan J, Clark D, Czodrowski P, et al. Applications of machine learning in drug discovery and development. Nat Rev Drug Discov. 2019; 18(6):463–477. DOI: https://doi.org/10.1038/s41573-019-0024-5.