Molecular Sets (MOSES): A Benchmarking Platform for Molecular Generation Models

By Daniil Polykovskiy, Alexander Zhebrak, Benjamín Sánchez-Lengeling, Sergey Golovanov, Oktai Tatanov, S. Belyaev, Rauf Kurbanov, A. Artamonov, V. Aladinskiy, M. Veselov and 4 more. Frontiers in Pharmacology, 2018

910 citations recorded by the source corpus.

Generative models are becoming a tool of choice for exploring the molecular space. These models learn on a large training dataset and produce novel molecular structures with similar properties. Generated structures can be utilized for virtual screening or training semi-supervized predictive models in the downstream tasks. While there are plenty of generative models, it is unclear how to compare and rank them. In this work, we introduce a benchmarking platform called Molecular Sets (MOSES) to standardize training and comparison of molecular generative models. MOSES provides training and testing datasets, and a set of metrics to evaluate the quality and diversity of generated structures. We have implemented and compared several molecular generation models and suggest to use our results as reference points for further advancements in generative chemistry research. The platform and source code are available at https://github.com/molecularsets/moses.

Subfields: Computational Theory and Mathematics, Materials Chemistry, Molecular Biology

View Paper: https://www.semanticscholar.org/paper/404571933e3e87942768c9a5cde8a6285732ad6f

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Theory and Mathematics

In Materials Chemistry

In Molecular Biology

Citations in the corpus

The Vendi Score: A Diversity Evaluation Metric for Machine Learning (2022, methodology): “We evaluate samples from generative models provided in the moses benchmark (Polykovskiy et al., 2020), using the ﬁrst 2500 valid molecules in each sample.”
GEOM: Energy-annotated molecular conformations for property prediction and molecular generation (2020, methodology): “Unlabeled molecular sets for generation tasks such as ChemBL [53], ZINC [54] and subsets [5], and related benchmarks are also reported [55, 56].”
Graph convolutional networks for computational drug development and discovery (2019, neutral): “[111] introduced MOSES for drug discovery (https://github.”
Hierarchical Generation of Molecular Graphs using Structural Motifs (2020, methodology): “To quantitatively evaluate the distance between two distributions, we compute Frechet distance between property distributions of molecules in the generated and test sets (Polykovskiy et al., 2018).”
GraphAF: a Flow-based Autoregressive Model for Molecular Graph Generation (2020, methodology): “QM9 contains 134k molecules with 9 heavy atoms, and MOSES is much larger and more challenging, which contains 1.9M molecules with up to 30 heavy atoms.”
Randomized SMILES strings improve the quality of molecular generative models (2019, methodology): “To prove that the molecules sampled from the randomized SMILES model are at least as diverse as those in the canonical, several physicochemical properties and metrics (as used in the MOSES benchmark [36]), such as molecular weight, logP, Synthetic Accessibility Score (SA) [37], Quantitative Estimate”
Artificial intelligence in drug discovery: recent advances and future perspectives (2021, methodology): “Recent efforts include the MOSES [154] and GuacaMol platforms [155], which implement several popular neural generative models, as well as more classical models ( e.g . genetic algorithms [156]), and provide several metrics for their comparison.”
DiGress: Discrete Denoising diffusion for graph generation (2022, methodology): “For MOSES, the reported scores for FCD, SNN, and Scaffold similarity are computed on the dataset made of separate scaffolds, which measures the ability of the networks to predict new ring structures.”