top of page

Topic: Learning 3D Representations of Molecular Chirality with Invariance to Bond Rotations.

Speaker: Keir Adams from Department of Chemical Engineering and the Center for Computational Science and Engineering at Massachusetts Institute of Technology (MIT)



We’re super delighted to announce our next speaker, Keir Adams, from Department of Chemical Engineering and the Center for Computational Science and Engineering at Massachusetts Institute of Technology (MIT) to bring us:


Learning 3D Representations of Molecular Chirality with Invariance to Bond Rotations


Abstract

Molecular stereochemistry strongly alters (bio)chemical interactions yet is often neglected in molecular deep learning. Tetrahedral (point) chirality, a form of stereochemistry describing relative spatial arrangements of bonded neighbors around tetrahedral carbon centers, especially influences substrate-catalyst binding and is therefore critical for pharmaceutical drug design and asymmetric catalysis. However, because chirality influences the set of 3D conformers accessible to the molecule without changing its 2D graph connectivity, modern graph neural networks (GNNs) designed for molecular property prediction often ignore chirality in their learned molecular representations. Most 2D GNNs at best use atomic labels to naïvely treat chirality, while E(3)-invariant 3D GNNs are invariant to chirality altogether.


To enable representation learning on molecules with defined chirality, we design an SE(3)-invariant neural network model that processes torsion angles of a 3D molecular conformer. We explicitly model conformational flexibility by integrating a novel type of invariance to rotations about internal molecular bonds directly into the neural architecture, mitigating the need for multi-conformer data augmentation. We showcase our model on four benchmarks: contrastive learning to distinguish conformers of different stereoisomers in a learned latent space, classification of chiral centers as R/S, prediction of how enantiomers rotate circularly polarized light, and ranking enantiomers by their docking scores in an enantiosensitive protein pocket. We compare our model, Chiral InterRoto Invariant Neural Network (ChIRo), with 2D and 3D GNNs to demonstrate that ChIRo achieves state of the art performance when learning chiral-sensitive functions from molecular structures. We envision that our model, alongside future work to explicitly learn E/Z isomerism and atropisomerism, will expand the capacity of deep learning to advance enantioselective reaction optimization, 3D drug design, and stereochemical property prediction.


Biography

Keir Adams is a 2nd year PhD student in the Department of Chemical Engineering and the Center for Computational Science and Engineering at MIT, advised by Prof. Connor W. Coley. He recently completed his undergraduate degree in chemistry and molecular engineering from the University of Chicago, where he performed research in computational astrochemistry and in renewable energy technologies. At MIT, Keir is generally interested in how 3D molecular representation learning can be used to help accelerate chemicals/materials design and discovery. He is especially interested in developing chemistry-tailored deep learning models that meaningfully understand both molecular stereochemistry and conformer flexibility. In his recent work, “Learning 3D Representations of Molecular Chirality with Invariance to Bond Rotations,” Keir addresses both these themes by designing a neural network that specially encodes 3D torsion angles to meaningfully learn tetrahedral chirality while also introducing a novel invariance to internal bond rotations directly into the model architecture.

Very much looking forward to the seminar – hope to see you all on Wednesday: (SGT) 9am, Wed, 24 November! https://mit.zoom.us/j/96231985116

Warmest regards, @Thway @kedar @Tonio


Topic: Deep Reasoning Networks in the landscape of automated materials discovery.

Speaker: Dr. John Gregoire from High Throughput Experimentation Group at Caltech



We’re super delighted to announce our next speaker, Dr. John Gregoire from High Throughput Experimentation Group at Caltech, to bring us:


Deep Reasoning Networks in the landscape of automated materials discovery


Abstract

Considering the confluence of hardware and software required for automating materials discovery, the aspects of research workflows that are particularly difficult to automate include the analysis of data in the context of deep prior knowledge. When the prior knowledge is data centric, supervised learning can be effectively utilized. In science, prior knowledge is more often rules or principles-centric, where reasoning about the data is more critical than pattern recognition. To realize the concept of model training under supervision by reasoning, the Gomes group (Cornell) developed Deep Reasoning Networks in collaboration with the Gregoire group (Caltech) to solve crystal structure phase mapping, i.e., the inference of phase diagrams from large quantities of x-ray diffraction data. The phase mapping problem provides an excellent platform for comparing complementary approaches to automated data interpretation, highlighting the benefits of artificial intelligence methods that integrate reasoning and learning.


Biography

Dr. John Gregoire is a Research Professor of Applied Physics and Materials Science and leads the High Throughput Experimentation group at Caltech. He is also the Team Lead for Photoactive Materials in the Liquid Sunlight Alliance (LiSA), a U.S. DOE Energy Innovation Hub. His research team explores, discovers, and understands energy-related materials via combinatorial and high throughput experimental methods and their integration with materials theory and artificial intelligence. The group seeks to accelerate scientific discovery by automating critical components of research workflows, from synthesis and screening to data interpretation and hypothesis generation. He received his BA in Math and Physics from Concordia College and PhD in Physics from Cornell University. For more information, please see https://gregoire.people.caltech.edu.

Very much looking forward to the seminar – hope to see you all on Wednesday: (SGT) 9am, Wed, 27 October! https://mit.zoom.us/j/96231985116

Warmest regards, @Thway @kedar @Tonio


Topic: Materials and Molecular Modeling, Imaging, Informatics and Integration.

Speaker: Prof. Seungbum Hong from Department of Materials Science and Engineering, KAIST, Korea



We’re super delighted to announce our next speaker, Prof. Seungbum Hong from Department of Materials Science and Engineering, KAIST, Korea, to bring us:


“Materials and Molecular Modeling, Imaging, Informatics and Integration”


Abstract

M3I3 is an algorithm to perform a reverse engineering of future materials. Fast followers usually copy the first movers’ products by reverse engineering them. For example, in case of the state-of-the-art battery products, the competitors dissect them into pieces and analyze the structure and composition of each part such as cathode, anode, electrolyte and separator. This so called “reverse engineering” is the cheapest way to catch up with the forefront runners in the ever-expanding competing world. For the front-runners, they also need a way to defend themselves and aggressively keep the distance from their competitors, and that’s why they invest a huge amount of resources into research and development of new materials, devices, systems and platforms, and file patents all over the world. M3I3 provides a means to achieve this goal effectively by mimicking “reverse engineering” strategy with a higher level of creativity. M3I3 reverse engineers future materials of interest with superior performance and reliability as well as with minimum cost and environmental impact.

How is this possible? Reverse engineering starts from analyzing the structure and composition of the cutting-edge materials or products. Once we determine the performance of our targeted future materials, we need to know the candidate structure and composition of the future materials. This knowledge can only be available if we know the structure-property or the property-structure relationship of all materials and molecules at all scales. High-quality multi-scale and multi-dimensional experimental data will the key to the success of our approach. But there are critical challenges such as collecting and analyzing those data with consistency. We hope to address those challenges and form a clearer idea for our future direction.


Biography

Prof. Seungbum Hong is currently a professor in the Department of Materials Science and Engineering at KAIST, Republic of Korea. He has received many honours and awards including Frontier Award, Samsung Advanced Institute of Technology (2007), Young Investigator Outstanding Achievement Award, International Symposium on Integrated Ferroelectrics (2008), Frontier Scientist (Physics), The Korean Academy of Science and Technology (KAST) (2014), Grand Prize for Learning & Teaching Innovation Awards (이수영 교수학습혁신대상), KAIST (2020), and KCA Excellent Expert Awards (KCA 우수전문인 어워즈), Korea Customer Appraisal (KCA), (2021). He has attended more than 190 conferences, published 160 papers and 5 books. The number of patent with his name currently stands at 79. His main research area of accomplishments includes energy harvesting, nanostructures, microscopy and much more.

Very much looking forward to the seminar – hope to see you all on Wednesday: (SGT) 9am, Wed, 13 October! https://mit.zoom.us/j/96231985116

Warmest regards, @Thway @kedar @Tonio


© 2024 by Kedar Hippalgaonkar. Created with Wix.com

Materials Science and Engineering, Nanyang Technological University

Institute of Materials Research and Engineering, Singapore

bottom of page