Congratulations to Our Researchers on Their Latest Publication in the Journal of Chemical Information and Modeling!
Congratulations to Dr. Pattipong Wisanpitayakorn, Dr. Kwanjeera Wanichthanarak, Dr. Narumol Jariyasopit, and Assoc. Prof. Dr. Sakda Khoomrung from SiCORE-MSB & SiMPC, Faculty of Medicine Siriraj Hospital, Mahidol University, on their latest publication in the Journal of Chemical Information and Modeling (September 2, 2026):
“Deep3DCCS: Geometry-Driven Collision Cross Section Prediction from Multi-View Molecular Projection Tensors”
What makes this work exciting?
The team developed Deep3DCCS, a deep learning model that learns molecular 3D shapes and predicts their behavior in ion mobility–mass spectrometry (IM-MS).
Rather than relying solely on conventional molecular descriptors, Deep3DCCS enables AI to “see” molecules from multiple perspectives and learn their three-dimensional shapes—much like image-recognition models learn to recognize objects from different views.
Deep3DCCS achieved prediction errors of approximately 3–4% on independent datasets, demonstrating its ability to capture meaningful molecular shape information for collision cross section prediction.
This approach could help fill gaps in molecular databases by predicting collision cross sections for compounds not yet experimentally characterized, supporting more efficient molecular identification and annotation in mass spectrometry-based research.
Read the full article:
https://pubs.acs.org/jcisd8/article/doi/10.1021/acs.jcim.6c00371/5425701/Deep3DCCS-Geometry-Driven-Collision-Cross-Section
News & Related Events
