Scientists and researchers today find themselves in a new era of drug discovery. Leveraging increasingly large chemical libraries, such as the latest ZINC22 database with over 37 billion molecules, offers opportunities for lead identification and optimization. Despite a large subset of these molecules being commercially available, such resources also pose challenges.
On one hand, researchers can swiftly tap into this expanding chemical space for therapeutic purposes. On the other hand, they face the daunting task of efficiently navigating libraries containing billions of molecules for drug discovery projects. This challenge is expected to persist as libraries inevitably continue to grow, complicating relatively simple tasks like identifying molecular analogues or performing in silico virtual screening studies against vast databases.
The emergence of machine learning and generative molecular design has led to the development of numerous startup companies and players in the drug discovery space. By using AI to optimize the multiparametric space of properties for drug-like molecules, they aim to expedite the journey from lead discovery to a preclinical candidate development. While these advances are promising, the conventional approach of using proprietary molecule libraries backed by validated experimental data will likely remain relevant.
Thus, there is a crucial need to develop novel algorithmic approaches to explore chemical space effectively while integrating data from experimental assays or computation to identify pertinent pharmacophores or moieties for specific molecular targets. Strategies to reduce computational costs while enhancing properties like lipophilicity, molecular weight, and H-bond donors and acceptors are essential. Recent innovations in this area include in silico combinatoric chemical fragment docking, machine learning approaches for predicting molecular properties, and active learning approaches for refining molecular predictions.
We invite researchers working in the fields of experimental and in silico drug design* and drug discovery to contribute to this research topic. We welcome the submission of Original Research, Reviews, or Perspective articles utilizing or discussing innovative methods for rapidly and efficiently exploring chemical space, comprising tens of billions of molecules, for therapeutic advances.
These approaches include, but are not limited to:
• Algorithms that are able to identify and enrich favorable areas of chemical space given a particular molecular target
• Algorithms that utilize machine learning or AI to optimize the exploration of chemical space and identify molecules with specific desired properties
• Algorithms that are able to utilize experimental data, molecular simulation data, or free-energy calculation data to inform the exploration of chemical space, and provide suggestions for further experimental optimization
• Algorithms that utilize graph-based representations or other approaches to efficiently perform similarity searches
• Algorithms that leverage high-performance or cloud-based resources to permit efficient exploration of chemical space and molecular properties
*Articles submitted to this collection will however be encouraged to incorporate experimental data used to validate the ensued in silico predictions.
Keywords:
chemical space, molecular design, virtual screening, molecular docking, cheminformatics, machine learning, drug discovery
Important Note:
All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.
Scientists and researchers today find themselves in a new era of drug discovery. Leveraging increasingly large chemical libraries, such as the latest ZINC22 database with over 37 billion molecules, offers opportunities for lead identification and optimization. Despite a large subset of these molecules being commercially available, such resources also pose challenges.
On one hand, researchers can swiftly tap into this expanding chemical space for therapeutic purposes. On the other hand, they face the daunting task of efficiently navigating libraries containing billions of molecules for drug discovery projects. This challenge is expected to persist as libraries inevitably continue to grow, complicating relatively simple tasks like identifying molecular analogues or performing in silico virtual screening studies against vast databases.
The emergence of machine learning and generative molecular design has led to the development of numerous startup companies and players in the drug discovery space. By using AI to optimize the multiparametric space of properties for drug-like molecules, they aim to expedite the journey from lead discovery to a preclinical candidate development. While these advances are promising, the conventional approach of using proprietary molecule libraries backed by validated experimental data will likely remain relevant.
Thus, there is a crucial need to develop novel algorithmic approaches to explore chemical space effectively while integrating data from experimental assays or computation to identify pertinent pharmacophores or moieties for specific molecular targets. Strategies to reduce computational costs while enhancing properties like lipophilicity, molecular weight, and H-bond donors and acceptors are essential. Recent innovations in this area include in silico combinatoric chemical fragment docking, machine learning approaches for predicting molecular properties, and active learning approaches for refining molecular predictions.
We invite researchers working in the fields of experimental and in silico drug design* and drug discovery to contribute to this research topic. We welcome the submission of Original Research, Reviews, or Perspective articles utilizing or discussing innovative methods for rapidly and efficiently exploring chemical space, comprising tens of billions of molecules, for therapeutic advances.
These approaches include, but are not limited to:
• Algorithms that are able to identify and enrich favorable areas of chemical space given a particular molecular target
• Algorithms that utilize machine learning or AI to optimize the exploration of chemical space and identify molecules with specific desired properties
• Algorithms that are able to utilize experimental data, molecular simulation data, or free-energy calculation data to inform the exploration of chemical space, and provide suggestions for further experimental optimization
• Algorithms that utilize graph-based representations or other approaches to efficiently perform similarity searches
• Algorithms that leverage high-performance or cloud-based resources to permit efficient exploration of chemical space and molecular properties
*Articles submitted to this collection will however be encouraged to incorporate experimental data used to validate the ensued in silico predictions.
Keywords:
chemical space, molecular design, virtual screening, molecular docking, cheminformatics, machine learning, drug discovery
Important Note:
All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.