More Like this

1. Deep Interactions for Multimodal Molecular Property Prediction

   https://doi.org/10.1007/978-981-96-8173-0_26  

   Soga, Patrick; Lei, Zhenyu; Bilodeau, Camille; Li, Jundong (June 2025, Springer Nature Singapore)

   Multi-modal learning by means of leveraging both 2D graph and 3D point cloud information has become a prevalent method to improve model performance in molecular property prediction. However, many recent techniques focus on specific pre-training tasks such as contrastive learning, feature blending, and atom/subgraph masking in order to learn multi-modality even though design of model architecture is also impactful for both pre-training and downstream task performance. Relying on pre-training tasks to align 2D and 3D modalities lacks direct interaction which may be more effective in multimodal learning. In this work, we propose MolInteract, which takes a simple yet effective architecture-focused approach to multimodal molecule learning which addresses these challenges. MolInteract leverages an interaction layer for fusing 2D and 3D information and fostering cross-modal alignment, showing strong results using even the simplest pre-training methods such as predicting features of the 3D point cloud and 2D graph. MolInteract exceeds state-of-the-art multimodal pre-training techniques and architectures on various downstream 2D and 3D molecule property prediction benchmark tasks. 

   more » « less
2. TorchMD: A Deep Learning Framework for Molecular Simulations

   https://doi.org/10.1021/acs.jctc.0c01343  

   Doerr, Stefan; Majewski, Maciej; Pérez, Adrià; Krämer, Andreas; Clementi, Cecilia; Noe, Frank; Giorgino, Toni; De Fabritiis, Gianni (April 2021, Journal of Chemical Theory and Computation)
3. Using Deep Learning to Identify Molecular Junction Characteristics

   https://doi.org/10.1021/acs.nanolett.0c00198  

   Fu, Tianren; Zang, Yaping; Zou, Qi; Nuckolls, Colin; Venkataraman, Latha (May 2020, Nano Letters)
4. Geometric Deep Learning for Molecular Crystal Structure Prediction

   https://doi.org/10.1021/acs.jctc.3c00031  

   Kilgour, Michael; Rogal, Jutta; Tuckerman, Mark (April 2023, Journal of Chemical Theory and Computation)
5. Deep Search for Molecular Oxygen in TW Hya

   https://doi.org/10.3847/1538-4357/acf717  

   Williams, Becky J.; Cleeves, L. Ilsedore; Eistrup, Christian; Ramsey, Jon P. (October 2023, The Astrophysical Journal)

   Abstract The dominant form of oxygen in cold molecular clouds is gas-phase carbon monoxide (CO) and ice-phase water (H₂O). Yet, in planet-forming disks around young stars, gas-phase CO and H₂O are less abundant relative to their interstellar medium values, and no other major oxygen-carrying molecules have been detected. Some astrochemical models predict that gas-phase molecular oxygen (O₂) should be a major carrier of volatile oxygen in disks. We report a deep search for emission from the isotopologue¹⁶O¹⁸O (N_J= 2₁− 0₁line at 233.946 GHz) in the nearby protoplanetary disk around TW Hya. We used imaging techniques and matched filtering to search for weak emission but do not detect¹⁶O¹⁸O. Based on our results, we calculate upper limits on the gas-phase O₂abundance in TW Hya of (6.4–70) × 10⁻⁷relative to H, which is 2–3 orders of magnitude below solar oxygen abundance. We conclude that gas-phase O₂is not a major oxygen carrier in TW Hya. Two other potential oxygen-carrying molecules, SO and SO₂, were covered in our observations, which we also do not detect. Additionally, we report a serendipitous detection of the C¹⁵NN_J= 2_5/2− 1_3/2hyperfine transitions,F= 3 − 2 andF= 2 − 1, at 219.9 GHz, which we found via matched filtering and confirm through imaging. 

   more » « less