Dataset Information

De Novo Design of Bioactive Small Molecules by Artificial Intelligence.

ABSTRACT: Generative artificial intelligence offers a fresh view on molecular design. We present the first-time prospective application of a deep learning model for designing new druglike compounds with desired activities. For this purpose, we trained a recurrent neural network to capture the constitution of a large set of known bioactive compounds represented as SMILES strings. By transfer learning, this general model was fine-tuned on recognizing retinoid X and peroxisome proliferator-activated receptor agonists. We synthesized five top-ranking compounds designed by the generative model. Four of the compounds revealed nanomolar to low-micromolar receptor modulatory activity in cell-based assays. Apparently, the computational model intrinsically captured relevant chemical and biological knowledge without the need for explicit rules. The results of this study advocate generative artificial intelligence for prospective de novo molecular design, and demonstrate the potential of these methods for future medicinal chemistry.

SUBMITTER: Merk D

PROVIDER: S-EPMC5838524 | biostudies-literature | 2018 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

De Novo Design of Bioactive Small Molecules by Artificial Intelligence.

Merk Daniel D Friedrich Lukas L Grisoni Francesca F Schneider Gisbert G

Molecular informatics 20180110 1-2

Generative artificial intelligence offers a fresh view on molecular design. We present the first-time prospective application of a deep learning model for designing new druglike compounds with desired activities. For this purpose, we trained a recurrent neural network to capture the constitution of a large set of known bioactive compounds represented as SMILES strings. By transfer learning, this general model was fine-tuned on recognizing retinoid X and peroxisome proliferator-activated receptor ...[more]

PMID: 29319225

Similar Datasets

Project description:Antimicrobial peptides (AMPs) provide protection against a variety of pathogenic bacteria and are, therefore, an important part of the innate immune system. Over the past decade, there has been considerable interest in developing AMPs as intravenously administered antibiotics. However, despite extensive efforts in the pharmaceutical and biotechnology industry, it has proven difficult to achieve this goal. While researchers have solved some relatively simple problems such as susceptibility to proteolysis, more severe problems have included the expense of the materials, toxicity, poor efficacy, and limited tissue distribution. In this Account, we describe our efforts to design and synthesize "foldamers"-- short sequence-specific oligomers based on arylamide and beta-amino acid backbones, which fold into well-defined secondary structures-- that could act as antimicrobial agents. We reasoned that small "foldamers" would be less expensive to produce than peptides, and might have better tissue distribution. It should be easier to fine-tune the structures and activities of these molecules to minimize toxicity. Because the activities of many AMPs depends primarily on their overall physicochemical properties rather than the fine details of their precise amino acid sequences, we have designed and synthesized very small "coarse-grained" molecules, which are far simpler than naturally produced AMPs. The molecular design of these foldamers epitomizes the positively charged amphiphilic structures believed to be responsible for the activity of AMPs. The designed oligomers show greater activity than the parent peptides. They have also provided leads for novel small molecule therapeutics that show excellent potency in animal models for multidrug resistant bacterial infections. In addition, such molecules can serve as relatively simple experimental systems for investigations aimed at understanding the mechanism of action for this class of antimicrobial agents. The foldamers' specificity for bacterial membranes relative to mammalian membranes appears to arise from differences in membrane composition and physical properties between these cell types. Furthermore, because experimental coarse-graining provided such outstanding results, we developed computational coarse-grained models to enable molecular dynamic simulations of these molecules with phospholipid membranes. These simulations allow investigation of larger systems for longer times than conventional molecular dynamics simulations, allowing us to investigate how physiologically relevant surface concentrations of AMP mimics affect the bilayer structure and properties. Finally, we apply the principles discovered through this work to the design of inexpensive antimicrobial polymers and materials.

Dataset Information

De Novo Design of Bioactive Small Molecules by Artificial Intelligence.

Publications

De Novo Design of Bioactive Small Molecules by Artificial Intelligence.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets