UROP Openings

Have a UROP opening you would like to submit?

Please fill out the form.

Submit your UROP opening

Creating chemistry-inspired atom-mapping library for chemical reactions (Remote)


Term:

Fall

Department:

3: Materials Science and Engineering

Faculty Supervisor:

Rafael Gomez-Bombarelli

Faculty email:

rafagb@mit.edu

Apply by:

9/15/2020

Contact:

angsj@mit.edu

Project Description

Atom-mapping between corresponding reactant(s) and product(s) molecules for a chemical reaction is often an important preprocessing step for many downstream cheminformatics operations. It is also essential for atoms to be correctly ordered before many autonomous transition-state search algorithms. In light of the group's interest in creating neural-network (NN) based reactive force fields, we aim to build an exhaustive chemistry-inspired python library for atom-mapping to facilitate automatic acquisition of datapoints for NN training. The student will first curate different classes of chemical reactions from organic chemistry textbooks, followed by breaking these reactions down into known elementary steps. Using the RDKit python package, we will then encode the atom-mapping for each of these elementary steps with SMILES and SMARTS. We envisage several significant downstream applications such as the prediction of stability of electrolyte molecules and the degradation of OLED materials.

Pre-requisites

Background in Python, Organic Chemistry, Reaction Mechanisms.