Collaborative Research: DMREF: Machine Learning and Robotics for the Data-Driven Design of Protein-polymer Hybrid Materials

合作研究：DMREF：用于蛋白质-聚合物杂化材料数据驱动设计的机器学习和机器人技术

• 批准号: 2118861
• 负责人: Michael Webb
• 金额: $ 48.41万
• 依托单位: Princeton University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-10-01 至 2025-09-30
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2118861&HistoricalAwards=false
• 关键词: Collaborative; Research; DMREF; Machine; Learning

Non-technical Description: Proteins are widely employed as agents for disease therapy and diagnosis, as well as catalytic components of commercial and industrial processes. In almost all applications, polymers are used as stabilizing constituents to increase durability of proteins in harsh and foreign environments, but the vast majority of stabilizing polymers provide modest protection due to non-specific interactions with the protein surface. Moreover, the surface properties of proteins used in these formulations are the result of evolution for working in mild, biological environments as opposed to the desired harsh, abiological ones. In principle, complex protein-polymer hybrids with tailored chemistries would facilitate superior protection by tightly wrapping polymer around the protein based on engineered complementary interactions. Such tailored formulations would stabilize the protein in its native state even under remarkably harsh conditions and have tremendous value in myriad industrial and military applications. However, these new materials are extraordinarily difficult to design due to their complexity. To address this challenge, this project will combine machine learning (ML) with robotics to rapidly discover new protein-polymer hybrid materials using data analytics and optimization tools. Over time, aggregated data will be used to train advanced ML models that can be applied to the prediction and design of functionality in a wide variety of novel materials. Equally important, this research will focus on the cross-disciplinary training of young data material scientists who will be prepared to enter the workforce and help revolutionize and engineer future materials. This research will feature a large collaborative effort between Rutgers University, Princeton University, and the Air Force Research Laboratory (AFRL).Technical Description: Current approaches to designing complementary protein-polymer interactions rely on labor intensive trial-and-error experimentation due to the lack a generalizable physicochemical framework that can guide the simultaneous design of both polymer and protein constituents at multiple length scales. This research will implement a novel machine learning-driven, bottom-up materials engineering paradigm for the design of protein-polymer hybrid particles; these hybrid particles will then be organized into protein-polymer hybrid assemblies to enhance stability in abiological environments. High-throughput polymer and protein production, characterization, multi-scale molecular simulation, and machine learning will be combined in a closed-loop fashion to both discover novel protein-polymer compositions and understand the physicochemical drivers for enhanced stability. Using these iterative Design-Build-Test-Learn cycles, underlying design principles for generating robust protein-polymer hybrids will be ascertained and the lead time for designing tailor-made protein-polymer hybrid materials will be substantially shortened.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

非技术描述：蛋白质被广泛用作疾病治疗和诊断的试剂，以及商业和工业过程的催化成分。在几乎所有的应用中，聚合物都被用作稳定组分，以提高蛋白质在恶劣和外来环境中的耐久性，但绝大多数稳定聚合物由于与蛋白质表面的非特异性相互作用而提供适度的保护。此外，这些配方中使用的蛋白质的表面特性是在温和的生物环境中工作的进化结果，而不是所需的苛刻非生物环境。原则上，具有量身定制的化学成分的复杂蛋白质-聚合物杂化材料将通过基于工程设计的互补相互作用将聚合物紧密包裹在蛋白质周围，从而提供更好的保护。这种量身定制的配方即使在非常恶劣的条件下也能稳定蛋白质的天然状态，在无数的工业和军事应用中具有巨大的价值。然而，这些新材料由于其复杂性，设计起来格外困难。为了应对这一挑战，该项目将把机器学习(ML)与机器人技术结合起来，利用数据分析和优化工具快速发现新的蛋白质-聚合物杂化材料。随着时间的推移，聚集的数据将被用于训练先进的ML模型，这些模型可以应用于各种新材料的功能预测和设计。同样重要的是，这项研究将侧重于对年轻的数据材料科学家进行跨学科培训，他们将准备进入劳动力大军，帮助革新和设计未来的材料。这项研究将以罗格斯大学、普林斯顿大学和空军研究实验室(AFRL)的大型合作为特色。技术描述：目前设计互补蛋白质-聚合物相互作用的方法依赖于劳动密集型反复试验，因为缺乏一个可通用的物理化学框架来指导在多个长度尺度上同时设计聚合物和蛋白质成分。这项研究将实施一种新的机器学习驱动的、自下而上的材料工程范式来设计蛋白质-聚合物杂化颗粒；然后这些杂化颗粒将被组织成蛋白质-聚合物杂化组装件，以提高在非生物环境中的稳定性。高通量聚合物和蛋白质的生产、表征、多尺度分子模拟和机器学习将以闭环方式结合在一起，以发现新的蛋白质-聚合物组成，并了解增强稳定性的物理化学驱动因素。使用这些迭代的设计-构建-测试-学习循环，将确定用于生成坚固的蛋白质-聚合物杂化材料的基本设计原则，并将大幅缩短设计定制蛋白质-聚合物杂化材料的前期时间。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Machine-Assisted Discovery of Chondroitinase ABC Complexes toward Sustained Neural Regeneration.

• DOI: 10.1002/adhm.202102101
• 发表时间: 2022-05
• 期刊: Advanced healthcare materials
• 影响因子: 10

Featurization strategies for polymer sequence or composition design by machine learning

通过机器学习进行聚合​​物序列或组合物设计的特征化策略

• DOI: 10.1039/d1me00160d
• 发表时间: 2022
• 期刊: Molecular Systems Design & Engineering
• 影响因子: 3.6
• 作者: Patel, Roshan A.;Borca, Carlos H.;Webb, Michael A.
• 通讯作者: Webb, Michael A.

Machine Learning on a Robotic Platform for the Design of Polymer-Protein Hybrids.

• DOI: 10.1002/adma.202201809
• 发表时间: 2022-07
• 期刊: ADVANCED MATERIALS
• 影响因子: 29.4
• 作者: Tamasi, Matthew J.;Patel, Roshan A.;Borca, Carlos H.;Kosuri, Shashank;Mugnier, Heloise;Upadhya, Rahul;Murthy, N. Sanjeeva;Webb, Michael A.;Gormley, Adam J.
• 通讯作者: Gormley, Adam J.