Peer-reviewedCC BYChemical Science
Synthesizability via reward engineering: expanding generative molecular design into synthetic space
4 affiliations
- 1Faculty of Pharmacy, University of Ljubljana, Ljubljana, Slovenia
- 2National Institute of Chemistry, Hajdrihova 19, Ljubljana 1000, Slovenia
- 3Molecular AI, Discovery Sciences, BioPharmaceuticals R&D, AstraZeneca, Gothenburg, Sweden
- 4Department of Physics, Chalmers University of Technology, Gothenburg, Sweden
Abstract
SynthSense couples generative molecular AI with retrosynthetic reward functions to produce novel, drug-like molecules that are not only feasible to synthesize but aligned with the parallel synthesis workflows of modern medicinal chemistry.
Related
- Molecular engineering of lysosome-based degraders unveils a rapidly expanding therapeutic strategy
- Synthetic Stochastic Time Series from Generative Models
- synthmc: a precedent-based synthesizability score for transition metal complexes
- Tissues As a Synthetic Target – A Framework For the Multiscale Molecular Design Of Thermogelling Polymers
- α‑Aminophosphonates in anticancer drug discovery: synthetic strategies, anticancer activities, structure–activity relationships and molecular mechanisms
BibTeX
@article{dekleva2026synthesizability,
title = {{Synthesizability via reward engineering: expanding generative molecular design into synthetic space}},
author = {Dominik Dekleva and Alexey Voronov and Jon Paul Janet and Albin Ekborg and Jure Borišek and Martina Hrast and Hannes H. Loeffler},
journal = {Chemical Science},
year = {2026},
month = may,
doi = {10.1039/d5sc09263a},
url = {https://doi.org/10.1039/d5sc09263a},
note = {CC BY},
}
RIS
TY - JOUR
TI - Synthesizability via reward engineering: expanding generative molecular design into synthetic space
AU - Dominik Dekleva
AU - Alexey Voronov
AU - Jon Paul Janet
AU - Albin Ekborg
AU - Jure Borišek
AU - Martina Hrast
AU - Hannes H. Loeffler
PY - 2026
DA - 2026/05/27
JO - Chemical Science
DO - 10.1039/d5sc09263a
UR - https://doi.org/10.1039/d5sc09263a
AB - SynthSense couples generative molecular AI with retrosynthetic reward functions to produce novel, drug-like molecules that are not only feasible to synthesize but aligned with the parallel synthesis workflows of modern medicinal chemistry.
SN - 2041-6520
SN - 2041-6539
ER -