Select Paper: ChainSpace: Sequence-Aware Design for Machine Learning-Driven Polymer Engineering

Research

Polymers have entered into every aspect of society, from familiar commodity items such as disposable plastic bags to high-value technological applications in electronics, energy, and medicine.

PhD Studies: University of Minnesota

In the modern drug discovery pharmaceutical pipeline, the vast majority of new drug candidates are extremely hydrophobic in nature, rendering them incompatible with the most popular, non-invasive, and low-cost route of administration for people worldwide, oral drug delivery.

To overcome this solubility problem, spray-dried dispersions (engineered solid-solid mixtures of amorphous drugs and polymeric carriers) have the potential of elevating drug solubility by orders of magnitude by maintaining drugs at supersaturation, thereby enhancing therapeutic potency and oral bioavailability over pharmaceutically-relevant timescales.

However, utilization of these materials is hindered by a lack molecular-level understanding between drugs and polymers. For my doctoral studies, rational approaches to establish structure-property relationships using well-defined, modular polymers were pursued.

Results elucidated new ways to carefully construct solid dispersion vehicles and controllably encapsulate drugs to direct storage and release properties. These findings provide essential synthetic tools and ingredients to meet unfulfilled drug formulation needs in today’s pharmaceutical landscape.

Senior Polymer Scientist: 3M

Advances in polymer science and engineering have enabled the opportunity to create a near-infinite spectrum of macromolecules. This synthetic versatility poses a new dilemma: how can decisions be made on what to create, so that the expanded diversity in building blocks can be harnessed into useful new products and technologies?

Polymer informatics (integration of machine learning/AI with experiment, simulation, and theory) has given glimpses into properties prediction for such heteropolymers. However, this emerging paradigm requires organized data libraries of polymer systems and formulations, with labeled features that are searchable.

I joined the 3M Corporate Research Laboratories at the end of 2018 as the first direct hire into the Materials Informatics Group, bringing an experimentalist's perspective to a team of software engineers and data scientists. Together we launched the 3M CRL MI Platform: a no-code, cloud-based system that lets researchers capture formulations, process parameters, and test results in one searchable place, with automated ingestion of routine characterization data (e.g., NMR and mechanical testing) and unique identifiers that link each sample's materials, processing, and properties for machine learning.

I led the development of high-throughput experimental capabilities to generate systematic datasets for the platform, whose initial focus was 3M's Adhesives Technology Platform (curable adhesives, epoxies, and pressure-sensitive adhesives). I also served as product owner and scrum leader for a $1.5M collaboration with Dr. Juan de Pablo's lab at UChicago combining polymer simulation, experiments, and machine learning. With Dr. Corinne Lipscomb, I co-authored an account of how we launched this initiative and the lessons learned for industrial MI programs: Ting & Lipscomb, Pure and Applied Chemistry (2022), https://doi.org/10.1515/pac-2022-0101

Dr. Corinne Lipscomb, Chief Product Owner of Materials Informatics, was featured in the ACS POLY Member Spotlight and discusses this further here.

Principal Scientist: Nanite

I was the first employee at Nanite, a Boston startup developing polymer nanoparticles for nonviral gene delivery. Nanite's SAYER™ platform connects high-throughput polymer chemistry, polyplex nanoparticle characterization, in vitro and in vivo screening, and predictive polymer design guided by machine learning and AI, so fit-for-purpose delivery vehicles can be rapidly produced and down-selected for cargo and tissue specificity.

My work focused on generating large-scale classes of polymer nanoparticles with precisely tailored chemical and physical attributes for controlled storage and release, feeding high-throughput in vitro to in vivo screening of gene delivery and AI-driven materials design. I led our team's perspective on this approach, covering polymer carriers for small molecules, nucleic acids, and proteins, and the digital infrastructure needed to make that data machine-actionable: Ting et al., Chemical Communications (2023), https://doi.org/10.1039/d3cc04705a

Google Sites

Report abuse