BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:"Machine learning in chemistry and beyond" (ChE-651) seminar by Ga
 be Gomes "Machine-Guided Catalyst Optimization"
DTSTART:20220927T153000
DTEND:20220927T163000
DTSTAMP:20260510T105800Z
UID:1a92f48c548bb28d99f793ab74ef09eeec5105587d3b138c50cd3d56
CATEGORIES:Conferences - Seminars
DESCRIPTION:Gabe is an assistant professor at the departments of Chemistry
  and Chemical Engineering at Carnegie Mellon University\, in Pittsburgh\, 
 USA. Gabe was born and raised in the countryside of the state of Rio de Ja
 neiro\, Brazil\, where he received his B.Sc. He earned his Ph.D. in Fall 2
 018 from Florida State University\, where he was awarded the LASER Fellows
 hip in 2014 and the 2016-2017 IBM Ph.D Scholarship. At FSU\, Gabe's resear
 ch was centered on the relationship between molecular structure and reacti
 vity\, focusing on the development and applications of stereoelectronic ef
 fects. For his work at FSU\, in 2018\, Gabe received several awards for hi
 s work in computational chemistry\, including his selection for the CAS Sc
 iFinder Future Leaders Program. In 2019\, Gabe joined the University of To
 ronto as a Postdoctoral Research Fellow in the Matter Lab\, led by Profess
 or Alán Aspuru-Guzik. In 2020\, Gabe was awarded the prestigious NSERC Ba
 nting Postdoctoral Fellowship with the project “Designing Catalysts with
  Artificial Intelligence\,” and has been featured on the “Next Great I
 mpossible” series by Merck/Milipore-Sigma. Gabe joined the Journal of Ch
 emical Information and Modeling as an Early Career Board member in 2021.\n
 \nThe ability to forge difficult chemical bonds through catalysis has tran
 sformed society on all fronts\, from feeding our ever-growing populations 
 to increasing our life-expectancies by synthesizing new drugs. Not only ha
 s the rise in popularity of metal-catalyzed cross-coupling reactions enabl
 ed us to make existing processes more efficient\, but it also has allowed 
 us to synthesize novel and unexplored molecules and materials\, unlocking 
 the technologies of the future.\n\nIn metal catalysis\, the choice of liga
 nd often leads to the most significant impact on the reaction outcomes\, s
 uch as yield or product selectivity. Identifying optimal metal-ligand comb
 inations can be a laborious experimental process. This practice is often h
 eld back by the difficulty of meaningfully comparing results with differen
 t ligands. I will introduce our efforts to develop a platform for inverse 
 design of catalysts utilizing high-throughput virtual screening and machin
 e learning (ML)\, coupled with an extensive ligands database. One of the m
 ost used strategies in ML for metal catalysis is the development of models
  based on reactant parameterization. Given that the space of potential lig
 ands is immense\, this tactic can be very challenging.\n\nWe have develope
 d a platform aimed at data-driven ligand optimization and their inverse de
 sign. At the center of this strategy is Kraken\,[1] a database of descript
 ors for organophosphorus ligands that encompasses features that are most i
 mportant for catalysis including conformational flexibility and ability fo
 r both coordinative and non-covalent bonding.\n\nHowever\, the development
  of ML solutions for chemical applications cannot be a simple application 
 of “out-of-the-shelf” methods. Molecular representation is at the core
  of chemical understanding. To date\, many ways to represent molecules for
  machines have been developed: different types of fingerprints\, language 
 model-based embeddings\, graph representations\, some physicochemical desc
 riptors. Still\, existing approaches lack quantum-mechanical information a
 bout molecular structure. The second part of this talk proposes a solution
  based on embedding stereoelectronic effects onto molecular graphs with gr
 aph neural networks.[2]\n\nWe developed an end-to-end pipeline that takes 
 a molecular geometry input and returns a representation with stereoelectro
 nic information. After enriching the standard molecular graphs with lone p
 airs (predicted by a separate model) and bond-orbital nodes\, we developed
  a multi-task approach that learns all the interaction values provided by 
 Natural Bond Orbital analysis. Finally\, we use the resulting representati
 ons for downstream tasks\, effectively displaying the approach’s vast ap
 plicability and increased performances.\n\n \n\nReferences\n\n[1]. T. Gen
 sch\, G. Gomes\, P. Friederich\, E. Peters\, T. Gaudin\, R. Pollice\, K. 
 Jorner\, A. Nigam\, M. L. D'Addario\, M. S. Sigman\, A. Aspuru-Guzik "A co
 mprehensive platform for the discovery and understanding of organophosphor
 us ligands in catalysis"\, J. Am. Chem. Soc. 2022\, 144 (3)\, 1205\n\n
 • webapp: kraken.cs.toronto.edu | • preprint on ChemRxiv\, 2021 | 
 • GitHub page  | • MORFEUS\n\n \n\n[2]. D. Boiko\, T. Reschützegge
 r\, B. Sanchez-Lengeling\, S. Blau\, G. Gomes* “Stereoelectronics-Awar
 e Molecular Representation Learning”\, ChemrXiv\, 2022\, doi: 10.26434/
 chemrxiv-2022-nz4pc\n\n \n\n\n 
LOCATION:https://epfl.zoom.us/j/68447908297?pwd=OU5JUGJUSUhZc0ZNYjQ2WENvYl
 NRdz09
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR