Simone Sciabola

Albert Bou, Morgan Thomas, Sebastian Dittert et al.

In recent years, reinforcement learning (RL) has emerged as a valuable tool in drug design, offering the potential to propose and optimize molecules with desired properties. However, striking a balance between capabilities, flexibility, reliability, and efficiency remains challenging due to the complexity of advanced RL algorithms and the significant reliance on specialized code. In this work, we introduce ACEGEN, a comprehensive and streamlined toolkit tailored for generative drug design, built using TorchRL, a modern RL library that offers thoroughly tested reusable components. We validate ACEGEN by benchmarking against other published generative modeling algorithms and show comparable or improved performance. We also show examples of ACEGEN applied in multiple drug discovery case studies. ACEGEN is accessible at \url{https://github.com/acellera/acegen-open} and available for use under the MIT license.

Wenlu Wang, Ye Wang, Honggang Zhao et al.

In the scope of drug discovery, the molecular design aims to identify novel compounds from the chemical space where the potential drug-like molecules are estimated to be in the order of 10^60 - 10^100. Since this search task is computationally intractable due to the unbounded search space, deep learning draws a lot of attention as a new way of generating unseen molecules. As we seek compounds with specific target proteins, we propose a Transformer-based deep model for de novo target-specific molecular design. The proposed method is capable of generating both drug-like compounds (without specified targets) and target-specific compounds. The latter are generated by enforcing different keys and values of the multi-head attention for each target. In this way, we allow the generation of SMILES strings to be conditional on the specified target. Experimental results demonstrate that our method is capable of generating both valid drug-like compounds and target-specific compounds. Moreover, the sampled compounds from conditional model largely occupy the real target-specific molecules' chemical space and also cover a significant fraction of novel compounds.

Pouya M. Ghari, Simone Sciabola, Ye Wang

Fine-tuning foundation models has emerged as a powerful approach for generating objects with specific desired properties. Reinforcement learning (RL) provides an effective framework for this purpose, enabling models to generate outputs that maximize a given reward function. However, in many applications such as text generation and drug discovery, it can be suboptimal to optimize using a single reward signal, as multiple evaluation criteria are often necessary. This paper proposes a novel reinforcement learning-based method for fine-tuning foundation models using multiple reward signals. By employing an iterative fine-tuning strategy across these rewards, our approach generalizes state-of-the-art RL-based methods. We further provide a theoretical analysis that offers insights into the performance of multi-reward RL fine-tuning. Experimental results across diverse domains including text, biological sequence, and small molecule generation, demonstrate the effectiveness of the proposed algorithm compared to state-of-the-art baselines.

Zhonglin Cao, Simone Sciabola, Ye Wang

Virtual screening of large compound libraries to identify potential hit candidates is one of the earliest steps in drug discovery. As the size of commercially available compound collections grows exponentially to the scale of billions, brute-force virtual screening using traditional tools such as docking becomes infeasible in terms of time and computational resources. Active learning and Bayesian optimization has recently been proven as effective methods of narrowing down the search space. An essential component in those methods is a surrogate machine learning model that is trained with a small subset of the library to predict the desired properties of compounds. Accurate model can achieve high sample efficiency by finding the most promising compounds with only a fraction of the whole library being virtually screened. In this study, we examined the performance of pretrained transformer-based language model and graph neural network in Bayesian optimization active learning framework. The best pretrained models identifies 58.97% of the top-50000 by docking score after screening only 0.6% of an ultra-large library containing 99.5 million compounds, improving 8% over previous state-of-the-art baseline. Through extensive benchmarks, we show that the superior performance of pretrained models persists in both structure-based and ligand-based drug discovery. Such model can serve as a boost to the accuracy and sample efficiency of active learning based molecule virtual screening.

Shiyun Wa, Yifei Wang, Simone Sciabola et al.

Molecular similarity plays a central role in ligand-based drug discovery, such as virtual screening, analog searching, and goal-directed molecular generation. However, traditional similarity measures, ranging from fingerprint-based Tanimoto coefficients to 3D shape overlays, are often computationally expensive at scale or rely on hand-crafted molecular descriptors. Meanwhile, many deep learning approaches to similarity-aware design still depend on similarity-specific supervision or costly data curation, limiting their generality across targets. In this work, we propose pretrained embedding distance (PED) as an effective alternative, computed directly from pretrained molecular models without task-specific training. Experimental results show that PED exhibits distinct correlations with traditional similarity metrics, and performs effectively in both ranking molecules for virtual screening and guiding molecular generation via reward design. These findings suggest that pretrained molecular embeddings capture rich structural information and can serve as a promising and scalable similarity measurement for modern AI-aided drug discovery.