Seminar: Polymer Informatics and High-Throughput Experimentation to Help Us Discover New Sustainable Polymers
Speaker: Bradley Olsen, Alexander and I. Michael Kasser Professor of Chemical Engineering, Massachusetts Institute of Technology
Host: Dan Schwartz
Tuesday, October 5, 2021 - 2:45 p.m. - JSCBB A108
Humanity is headed for a series of sustainability crises, and our pace of innovation and technology transition must substantially accelerate to avoid their worst effects. In particular, our widespread adoption of polymers, which has enabled an era of unprecedented increase in the global standard of living due to clean water, uncontaminated food, and wider provision of health care, has led to a waste crisis of daunting proportions that is accelerating exponentially with our growing polymer production. Addressing this issue without compromising on the societal benefits that polymers bring urgently demands the discovery of materials.
In speaking to many broad constituencies in the polymer development pipeline, we learned that the largest roadblocks to faster innovation are the friction and barriers in our knowledge sharing ecosystem. Despite these well-known issues, polymer science has lagged substantially behind other branches of chemistry in applying informatics tools because polymers present several unique challenges for data science: (1) stochastic structures, (2) small and disparate data, (3) challenging nomenclature, and (4) multi-scale chemistry and physics. To address these issues in data sharing, we have invented new representations for polymer chemical structure formulas (BigSMILES), data schemas that allow data to be organized in a way that preserves the structure of the way it was generated (PolyDAT and CRIPT), and a new polymer chemical structure search language (BigSMARTS). These tools are being made publicly available through our non-profit CRIPT project which is developing and maintaining a FAIR database for polymer data.
We are realizing the promise of these tools as drivers for our own high-throughput experimentation, exploiting them to facilitate the development of quantitative structure-property relationship (QSPR) predictions for degradability in polymers. Using parallel batch synthesis, we have prepared a library of over 600 different polyesters from over 100 unique monomers representing all major routes to polyesterification and a diverse set of heteroatom functionalities. We have then adapted the clear zone assay for bacterial screening into a high-throughput assay for polymer biodegradation, allowing the entire library to be screened in a mater of months. Using our digital tools to ingest and organize the data, we can then apply different models for prediction of degradation rates based on chemical structure. Importantly, the open format of this project allows us to share and merge our data with others, synergizing efforts to improve the predictive capability of the models. The goal is that this technique can be used in order to predict the biodegradability of polymers synthesized using proposed new monomers derived from biosynthetic pathways even before they are prepared.
Bradley Olsen is the Alexander and I. Michael Kasser Professor of Chemical Engineering at MIT. He earned his S.B. in Chemical Engineering at MIT, his Ph.D. in Chemical Engineering at the University of California – Berkeley, and was a postdoctoral scholar at the California Institute of Technology. He started as an assistant professor at MIT in December 2009. Olsen’s research expertise is in materials chemistry and polymer physics, with a particular emphasis on molecular self-assembly, protein materials, polymer networks, and polymer informatics. He is a fellow of the ACS and member of APS and AIChE.