Multi-objective synthesis planning by means of Monte Carlo Tree search

We introduce a multi-objective search algorithm for retrosynthesis planning, based on a Monte Carlo Tree search formalism. The multi-objective search allows for combining diverse set of objectives without considering their scale or weighting factors. To benchmark this novel algorithm, we employ four...

Full description

Saved in:
Bibliographic Details
Main Authors: Helen Lai, Christos Kannas, Alan Kai Hassen, Emma Granqvist, Annie M. Westerlund, Djork-Arné Clevert, Mike Preuss, Samuel Genheden
Format: Article
Language:English
Published: Elsevier 2025-06-01
Series:Artificial Intelligence in the Life Sciences
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2667318525000066
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850128727263936512
author Helen Lai
Christos Kannas
Alan Kai Hassen
Emma Granqvist
Annie M. Westerlund
Djork-Arné Clevert
Mike Preuss
Samuel Genheden
author_facet Helen Lai
Christos Kannas
Alan Kai Hassen
Emma Granqvist
Annie M. Westerlund
Djork-Arné Clevert
Mike Preuss
Samuel Genheden
author_sort Helen Lai
collection DOAJ
description We introduce a multi-objective search algorithm for retrosynthesis planning, based on a Monte Carlo Tree search formalism. The multi-objective search allows for combining diverse set of objectives without considering their scale or weighting factors. To benchmark this novel algorithm, we employ four objectives in a total of eight retrosynthesis experiments on a PaRoutes benchmark set. The objectives range from simple ones based on starting material and step count to complex ones based on synthesis complexity and route similarity. We show that with the careful employment of complex objectives, the multi-objective algorithm can outperform the single-objective search and provides a more diverse set of solutions. However, for many target compounds, the single- and multi-objective settings are equivalent. Nevertheless, our algorithm provides a framework for incorporating novel objectives for specific applications in synthesis planning.
format Article
id doaj-art-1938e87f3454471fba17dfb16ab3f144
institution OA Journals
issn 2667-3185
language English
publishDate 2025-06-01
publisher Elsevier
record_format Article
series Artificial Intelligence in the Life Sciences
spelling doaj-art-1938e87f3454471fba17dfb16ab3f1442025-08-20T02:33:12ZengElsevierArtificial Intelligence in the Life Sciences2667-31852025-06-01710013010.1016/j.ailsci.2025.100130Multi-objective synthesis planning by means of Monte Carlo Tree searchHelen Lai0Christos Kannas1Alan Kai Hassen2Emma Granqvist3Annie M. Westerlund4Djork-Arné Clevert5Mike Preuss6Samuel Genheden7Molecular AI, Discovery Sciences, R&D, AstraZeneca, Cambridge, UK; Corresponding author.Molecular AI, Discovery Sciences, R&D, AstraZeneca, Gothenburg, SwedenLeiden Institute of Advanced Computer Science, Leiden University, Leiden, The Netherlands; Machine Learning Research, Pfizer Research and Development, Berlin, GermanyMolecular AI, Discovery Sciences, R&D, AstraZeneca, Gothenburg, Sweden; Department of Computer Science and Engineering, Chalmers University of Technology, Gothenburg, SwedenMolecular AI, Discovery Sciences, R&D, AstraZeneca, Gothenburg, SwedenMachine Learning Research, Pfizer Research and Development, Berlin, GermanyLeiden Institute of Advanced Computer Science, Leiden University, Leiden, The NetherlandsMolecular AI, Discovery Sciences, R&D, AstraZeneca, Gothenburg, SwedenWe introduce a multi-objective search algorithm for retrosynthesis planning, based on a Monte Carlo Tree search formalism. The multi-objective search allows for combining diverse set of objectives without considering their scale or weighting factors. To benchmark this novel algorithm, we employ four objectives in a total of eight retrosynthesis experiments on a PaRoutes benchmark set. The objectives range from simple ones based on starting material and step count to complex ones based on synthesis complexity and route similarity. We show that with the careful employment of complex objectives, the multi-objective algorithm can outperform the single-objective search and provides a more diverse set of solutions. However, for many target compounds, the single- and multi-objective settings are equivalent. Nevertheless, our algorithm provides a framework for incorporating novel objectives for specific applications in synthesis planning.http://www.sciencedirect.com/science/article/pii/S2667318525000066Monte Carlo Tree searchPareto optimalityMulti-objective optimizationReinforcement learningMarkov Decision ProcessPolicy network
spellingShingle Helen Lai
Christos Kannas
Alan Kai Hassen
Emma Granqvist
Annie M. Westerlund
Djork-Arné Clevert
Mike Preuss
Samuel Genheden
Multi-objective synthesis planning by means of Monte Carlo Tree search
Artificial Intelligence in the Life Sciences
Monte Carlo Tree search
Pareto optimality
Multi-objective optimization
Reinforcement learning
Markov Decision Process
Policy network
title Multi-objective synthesis planning by means of Monte Carlo Tree search
title_full Multi-objective synthesis planning by means of Monte Carlo Tree search
title_fullStr Multi-objective synthesis planning by means of Monte Carlo Tree search
title_full_unstemmed Multi-objective synthesis planning by means of Monte Carlo Tree search
title_short Multi-objective synthesis planning by means of Monte Carlo Tree search
title_sort multi objective synthesis planning by means of monte carlo tree search
topic Monte Carlo Tree search
Pareto optimality
Multi-objective optimization
Reinforcement learning
Markov Decision Process
Policy network
url http://www.sciencedirect.com/science/article/pii/S2667318525000066
work_keys_str_mv AT helenlai multiobjectivesynthesisplanningbymeansofmontecarlotreesearch
AT christoskannas multiobjectivesynthesisplanningbymeansofmontecarlotreesearch
AT alankaihassen multiobjectivesynthesisplanningbymeansofmontecarlotreesearch
AT emmagranqvist multiobjectivesynthesisplanningbymeansofmontecarlotreesearch
AT anniemwesterlund multiobjectivesynthesisplanningbymeansofmontecarlotreesearch
AT djorkarneclevert multiobjectivesynthesisplanningbymeansofmontecarlotreesearch
AT mikepreuss multiobjectivesynthesisplanningbymeansofmontecarlotreesearch
AT samuelgenheden multiobjectivesynthesisplanningbymeansofmontecarlotreesearch