Abstract:In structurally constrained molecular optimization, state-of-the-art methods restart an expensive oracle-driven search from scratch for every new input structure, scaling poorly to settings with many starting structures or expensive oracles. While amortized approaches that learn a transferable policy could in principle remove this bottleneck, existing methods struggle to generalize to diverse structural constraints at inference time. We present AMORTIX, an amortized Graph Transformer model that natively supports such constraints, optimizing molecular structures in a single forward pass with zero inference-time oracle calls. A central challenge for amortized training in this domain is that optimization difficulty varies drastically across starting structures. We show that, under this heterogeneity, standard reinforcement learning methods fail to stabilize training, and address this by normalizing rewards within groups of completions sharing the same starting structure. We evaluate on structurally constrained single- and multi-target kinase inhibitor design, and on a few-shot prodrug case study. AMORTIX outperforms both amortized and instance-optimization baselines on goal-directed scaffold decoration and ranks first among amortized methods on the PMO benchmark; the prodrug case study further demonstrates transfer of a learned modification rule to unseen drug structures. Code is available at this https URL.
| Subjects: | Machine Learning (cs.LG) |
| ACM classes: | I.2.6; I.2.1 |
| Cite as: | arXiv:2602.12162 [cs.LG] |
| (or arXiv:2602.12162v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2602.12162 arXiv-issued DOI via DataCite |
From: Muhammad Bin Javaid [view email]
[v1]
Thu, 12 Feb 2026 16:43:59 UTC (3,372 KB)
[v2]
Thu, 19 Feb 2026 23:41:49 UTC (3,372 KB)
[v3]
Fri, 8 May 2026 01:26:51 UTC (3,099 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。