The paper evaluates molecular representations used in cheminformatics and small‑molecule machine learning, such as fingerprints and learned embeddings. It finds that current state‑of‑the‑art embedding models do not outperform traditional fingerprints or physicochemical descriptors in capturing graph structure, 3D shape, or ADMET similarity. Additionally, fingerprints exhibit significant limitations, with 225 million to 4 billion molecules in the 10.8 billion‑molecule Enamine REAL database sharing non‑unique representations, which impacts virtual screening.
arXiv:2608. 05336v1 Announce Type: cross Abstract: Molecular representations are essential for the evaluation of molecular similarity and the development of structure-property relationships.
arXiv:2604. 27810v2 Announce Type: replace Abstract: Computational molecular representations underpin virtual screening, property prediction, and materials discovery.
arXiv:2604. 24474v2 Announce Type: replace Abstract: Molecular similarity plays a central role in ligand-based drug discovery, such as virtual screening, analog searching, and goal-directed molecular generation.
The study benchmarked 28 pretrained molecular embedding models—including graph neural networks, text transformers, and hybrids—across 25 classification datasets using a controlled framework with hierarchical Bayesian Bradley‑Terry testing. Except for CheMeleon, none of the neural models surpassed the classical ECFP count fingerprint, and external validation with scaffold cross‑validation and Bayesian hierarchical t‑tests found no clear advantage for any embedding method. The authors recommend evaluating both ECFP fingerprints and CheMeleon embeddings in new QSAR/QSPR studies, emphasizing the need for rigorous evaluation standards in molecular representation learning.
arXiv:2609.15611v1 Announce Type: cross Abstract: Molecular property prediction requires representations that generalize from limited labeled data to structurally novel compounds. Existing molecular...
The paper investigates how explicitly supervising molecular embeddings with a molecule’s Bemis‑Murcko scaffold influences representation learning. Experiments compare Euclidean and Lorentz contrastive objectives under two augmentation strengths, showing that scaffold‑supervised models consistently group molecules by identical and related scaffolds. These embeddings also enhance property prediction on several tasks, though the magnitude of improvement varies with the target property and the geometry used.
The study evaluates four pretrained molecular language models on six virtual libraries covering drug discovery, organic materials, and catalysis. It finds that native embeddings vary widely in performance, while molecular fingerprints remain consistently strong. Fine‑tuning the models on library‑specific data markedly improves sample efficiency, with several adapted encoders outperforming others across all tasks.
arXiv:2510.07289v2 Announce Type: replace Abstract: Molecular graph representation learning is widely used in chemical and biomedical research. While pre-trained 2D graph encoders have demonstrated s...
arXiv:2510. 14217v2 Announce Type: replace Abstract: The spectral properties of feature embeddings offer critical insights into model generalization and representation quality.
WEECFP-SuRGE introduces a position‑aware substructure encoding method that combines tokenized hierarchical Morgan fingerprints with graph‑distance‑dependent rotations applied at the input and within transformer self‑attention. The approach captures local chemistry, long‑range interactions, and molecular topology without requiring external pretraining or 3‑D conformer generation. Benchmarks on MoleculeNet and the Therapeutic Data Commons ADMET datasets show competitive performance, and a reconstruction procedure correctly identifies constitutional isomers for 92.6% of a 4,200‑molecule library.
arXiv:2607. 05736v1 Announce Type: new Abstract: Molecular property prediction often relies on isolated data modalities, where continuous 3D graph neural networks (GNNs) struggle to efficiently capture long-range topological dependencies and exact macroscopic heuristics.
arXiv:2605. 16823v2 Announce Type: replace Abstract: Large language models succeed by combining large-scale pretraining with meaningful discrete tokens.