Skip to content
DraconiticResearch
Search Sources API
197,768 studies31 sourcesUpdated 40 min ago
PreprintCC BY-NC-NDChemRxiv

From Computational Chemistry to Generative Models: A Survey of AI-Driven Small-Molecule Drug Discovery

  • Houman Kazemzadeh1,2
  • Kiarash Mokhtari3,4
  • Seyed Reza Tavakoli5,6
  • Nazanin Mirzaei5,6
  • Ali Reza Keivanimehr7,8
  • Shayan Majidifar9,6
  • Ali Behrad10,6
  • Farbod Davoodi11,12
  • Sina Khosravi13,14
  • Behdad Alikhani1,2
  • Ali Sabzi9,6
  • Parham Abed Azad9,6
8 more authors
  • Arshia Abolghasemi15,16
  • Afshin Cheraghzade17,18
  • Siavash Ahmadi19,6
  • Babak Khalaj10,6
  • Mohammad Hossein Rohban9,6
  • Gholamali Aminian20
  • Ali Baheri
  • Koorosh Aslansefat21,22
22 affiliations
  1. 1Department of Medicinal Chemistry, Faculty of Pharmacy
  2. 2Tehran University of Medical Sciences
  3. 3Faculty IV–Electrical Engineering and Computer Science
  4. 4Technische Universität Berlin
  5. 5Department of Mathematical Sciences
  6. 6Sharif University of Technology
  7. 7Department of Management, Science and Technology
  8. 8Amirkabir University of Technology (Tehran Polytechnic)
  9. 9Department of Computer Engineering
  10. 10Department of Electrical Engineering
  11. 11Department of Computer Science
  12. 12Missouri University of Science and Technology
  13. 13Department of Medicinal Chemistry, School of Pharmacy
  14. 14Mashhad University of Medical Sciences
  15. 15School of Electrical and Computer Engineering
  16. 16University College of Engineering, University of Tehran
  17. 17Department of Occupational Health Engineering, School of Health
  18. 18Semnan University of Medical Sciences
  19. 19Electronics Research Institute
  20. 20The Alan Turing Institute
  21. 21Computer Science, School of Physical and Digital Sciences, Faculty of Science and Engineering
  22. 22University of Hull
Open at sourcePDFDOICite

Abstract

Small-molecule drug discovery requires navigating enormous chemical space to identify candidates that are simultaneously potent, selective, synthetically accessible, and acceptable across pharmacokinetic and safety profiles; modern generative AI extends a long computational medicinal chemistry lineage that began with QSAR, pharmacophore modeling, and early structure-based de novo design. This survey focuses primarily on developments from 2017 to 2025, with the literature updated through 31 January 2026. Using a medicinal-chemistry lens, it first establishes the shared machinery of the field: molecular representations such as SMILES, SELFIES, graphs, fingerprints, three-dimensional coordinates, voxels, and latent vectors; commonly used datasets and benchmarks; and graph neural network encoders, including message-passing, directional, and equivariant architectures. We compare five principal generative mechanisms—autoregressive models, variational autoencoders, generative adversarial networks, normalizing flows, and diffusion models—together with reinforcement learning as a major goal-directed optimization framework. Across these approaches, we show that current methods offer complementary rather than interchangeable trade-offs, and we highlight recurring challenges, including validity versus synthesizability, limited prospective experimental validation, mode collapse and diversity loss, stereochemical and physical realism in three-dimensional design, computational cost, data scarcity, and the emerging convergence toward scaffold-conditioned generation, hybrid architectures, developability-aware optimization, and molecular foundation models.

Related

  1. Quantifying Multi-Objective Optimization in Generative Chemistry for Molecule Design
    Paula Torren-Peraire, Nadine Schneider, Raquel Rodríguez-Pérez +2
    ChemRxivPreprintCC BY-NC-NDFull textGeneral chemistryChemRxiv
  2. Single-molecule cold chemistry: Isotopic branching ratios of the H2O+ + HD reaction at cryogenic temperatures
    Nanditha Sunil Kumar, Prerna Paliwal, Mikhail Popov +4
    arXivPreprintCC BY 4.0Full textPhysical chemistryarXiv
  3. Built in Minutes, Checked by Script: Instructor-Made Interactive Chemistry Study Guides with Generative AI
    John Picione
    ChemRxivPreprintLicence not statedGeneral chemistryChemRxiv
  4. Escape Rooms, Arcades, and a Grain of Salt: Five Original Games for Introductory Chemistry, Built by One Instructor with Generative AI
    John Picione
    ChemRxivPreprintLicence not statedGeneral chemistryChemRxiv
  5. Robust generative transition-state models for unseen chemistry
    Samir Darouich, Jacob Toney, Weiliang Luo +3
    Nature Computational SciencePeer-reviewedLicence not statedFull text11 citedMaterials & polymersOpenAlex

Cite

BibTeXRIS
BibTeX
@misc{kazemzadeh2026from,
  title = {{From Computational Chemistry to Generative Models: A Survey of AI-Driven Small-Molecule Drug Discovery}},
  author = {Houman Kazemzadeh and Kiarash Mokhtari and Seyed Reza Tavakoli and Nazanin Mirzaei and Ali Reza Keivanimehr and Shayan Majidifar and Ali Behrad and Farbod Davoodi and Sina Khosravi and Behdad Alikhani and Ali Sabzi and Parham Abed Azad and Arshia Abolghasemi and Afshin Cheraghzade and Siavash Ahmadi and Babak Khalaj and Mohammad Hossein Rohban and Gholamali Aminian and Ali Baheri and Koorosh Aslansefat},
  howpublished = {ChemRxiv},
  year = {2026},
  month = sep,
  doi = {10.26434/chemrxiv.15008163/v1},
  url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008163/v1},
  note = {CC BY-NC-ND},
}
RIS
TY  - UNPB
TI  - From Computational Chemistry to Generative Models: A Survey of AI-Driven Small-Molecule Drug Discovery
AU  - Houman Kazemzadeh
AU  - Kiarash Mokhtari
AU  - Seyed Reza Tavakoli
AU  - Nazanin Mirzaei
AU  - Ali Reza Keivanimehr
AU  - Shayan Majidifar
AU  - Ali Behrad
AU  - Farbod Davoodi
AU  - Sina Khosravi
AU  - Behdad Alikhani
AU  - Ali Sabzi
AU  - Parham Abed Azad
AU  - Arshia Abolghasemi
AU  - Afshin Cheraghzade
AU  - Siavash Ahmadi
AU  - Babak Khalaj
AU  - Mohammad Hossein Rohban
AU  - Gholamali Aminian
AU  - Ali Baheri
AU  - Koorosh Aslansefat
PY  - 2026
DA  - 2026/09/01
PB  - ChemRxiv
DO  - 10.26434/chemrxiv.15008163/v1
UR  - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008163/v1
AB  - Small-molecule drug discovery requires navigating enormous chemical space to identify candidates that are simultaneously potent, selective, synthetically accessible, and acceptable across pharmacokinetic and safety profiles; modern generative AI extends a long computational medicinal chemistry lineage that began with QSAR, pharmacophore modeling, and early structure-based de novo design. This survey focuses primarily on developments from 2017 to 2025, with the literature updated through 31 January 2026. Using a medicinal-chemistry lens, it first establishes the shared machinery of the field: molecular representations such as SMILES, SELFIES, graphs, fingerprints, three-dimensional coordinates, voxels, and latent vectors; commonly used datasets and benchmarks; and graph neural network encoders, including message-passing, directional, and equivariant architectures. We compare five principal generative mechanisms—autoregressive models, variational autoencoders, generative adversarial networks, normalizing flows, and diffusion models—together with reinforcement learning as a major goal-directed optimization framework. Across these approaches, we show that current methods offer complementary rather than interchangeable trade-offs, and we highlight recurring challenges, including validity versus synthesizability, limited prospective experimental validation, mode collapse and diversity loss, stereochemical and physical realism in three-dimensional design, computational cost, data scarcity, and the emerging convergence toward scaffold-conditioned generation, hybrid architectures, developability-aware optimization, and molecular foundation models.
ER  -