Beam search for automated design and scoring of novel ROR ligands with machine intelligence

Angewandte Chemie ◽

10.1002/ange.202104405 ◽

2021 ◽

Author(s):

Michael Moret ◽

Moritz Helmstädter ◽

Francesca Grisoni ◽

Gisbert Schneider ◽

Daniel Merk

Keyword(s):

Automated Design ◽

Machine Intelligence ◽

Beam Search

Download Full-text

Beam Search Sampling for Molecular Design and Intrinsic Prioritization with Machine Intelligence

10.26434/chemrxiv.14153408 ◽

2021 ◽

Author(s):

Michael Moret ◽

Moritz Helmstädter ◽

Francesca Grisoni ◽

Gisbert Schneider ◽

Daniel Merk

Keyword(s):

De Novo ◽

Molecular Design ◽

Search Algorithm ◽

Sampling Technique ◽

Machine Intelligence ◽

Language Models ◽

Scoring Functions ◽

Beam Search ◽

De Novo Drug Design ◽

Chemical Language

Chemical language models enable de novo drug design without the requirement for explicit molecular construction rules. While such models have been applied to generate novel compounds with desired bioactivity, the actual prioritization and selection of the most promising computational designs remains challenging. In this work, we leveraged the probabilities learnt by chemical language models with the beam search algorithm as a model-intrinsic technique for automated molecule design and scoring. Prospective application of this method yielded three novel inverse agonists of retinoic acid receptor-related orphan receptors (RORs). Each design was synthesizable in three reaction steps and presented low-micromolar to nanomolar potency towards RORg. This model-intrinsic sampling technique eliminates the strict need for external compound scoring functions, thereby further extending the applicability of generative artificial intelligence to data-driven drug discovery.<br>

Download Full-text

Beam Search Sampling for Molecular Design and Intrinsic Prioritization with Machine Intelligence

10.26434/chemrxiv.14153408.v1 ◽

2021 ◽

Author(s):

Michael Moret ◽

Moritz Helmstädter ◽

Francesca Grisoni ◽

Gisbert Schneider ◽

Daniel Merk

Keyword(s):

De Novo ◽

Molecular Design ◽

Search Algorithm ◽

Sampling Technique ◽

Machine Intelligence ◽

Language Models ◽

Scoring Functions ◽

Beam Search ◽

De Novo Drug Design ◽

Chemical Language

Chemical language models enable de novo drug design without the requirement for explicit molecular construction rules. While such models have been applied to generate novel compounds with desired bioactivity, the actual prioritization and selection of the most promising computational designs remains challenging. In this work, we leveraged the probabilities learnt by chemical language models with the beam search algorithm as a model-intrinsic technique for automated molecule design and scoring. Prospective application of this method yielded three novel inverse agonists of retinoic acid receptor-related orphan receptors (RORs). Each design was synthesizable in three reaction steps and presented low-micromolar to nanomolar potency towards RORg. This model-intrinsic sampling technique eliminates the strict need for external compound scoring functions, thereby further extending the applicability of generative artificial intelligence to data-driven drug discovery.<br>

Download Full-text

Human-competitive machine invention by means of genetic programming

Artificial intelligence for engineering design analysis and manufacturing ◽

10.1017/s0890060408000127 ◽

2008 ◽

Vol 22 (3) ◽

pp. 185-193 ◽

Cited By ~ 6

Author(s):

John R. Koza

Keyword(s):

Developmental Biology ◽

Natural Selection ◽

Gene Duplication ◽

Genetic Programming ◽

Gene Deletion ◽

Automated Design ◽

Machine Intelligence ◽

Systematic Method

AbstractGenetic programming is a systematic method for getting computers to automatically solve problems. Genetic programming uses the Darwinian principle of natural selection and analogs of recombination (crossover), mutation, gene duplication, gene deletion, and certain mechanisms of developmental biology to progressively breed, over a series of many generations, an improved population of candidate solutions to a problem. This paper makes the points that genetic programming now routinely delivers human-competitive machine intelligence for problems of automated design and can serve as an automated invention machine.

Download Full-text