Chemical and biological language models in molecular design: opportunities, risks and scientific reasoning
In the physical and life sciences, including drug discovery, the use of deep learning (DL) models, such as language models (LMs) or graph neural networks (GNNs), is on the rise for various applications. However, the versatility of DL architectures comes at a price. While they open the door to novel applications, their use is also prone to misconceptions or misunderstandings, often leading to false assumptions and controversial views of scientific applications. This commentary discusses the general requirements, potential caveats or pitfalls and explanation of DL models in the context of molecular design
- Published in:
Future Science OA - Type:
Article - Authors:
Bajorath, Jürgen - Year:
2024
Citation information
Bajorath, Jürgen: Chemical and biological language models in molecular design: opportunities, risks and scientific reasoning, Future Science OA, 2024, https://www.future-science.com/doi/10.2144/fsoa-2023-0318, Bajorath.2024b,
@Article{Bajorath.2024b,
author={Bajorath, Jürgen},
title={Chemical and biological language models in molecular design: opportunities, risks and scientific reasoning},
journal={Future Science OA},
url={https://www.future-science.com/doi/10.2144/fsoa-2023-0318},
year={2024},
abstract={In the physical and life sciences, including drug discovery, the use of deep learning (DL) models, such as language models (LMs) or graph neural networks (GNNs), is on the rise for various applications. However, the versatility of DL architectures comes at a price. While they open the door to novel applications, their use is also prone to misconceptions or misunderstandings, often leading to...}}
