CAREER: Teaching Machines to Design Self-Assembling Materials

职业：教授机器设计自组装材料

• 批准号: 1350008
• 负责人: Andrew Ferguson
• 金额: $ 45万
• 依托单位: University of Illinois at Urbana-Champaign
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2014
• 资助国家: 美国
• 起止时间: 2014-06-01 至 2018-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1350008&HistoricalAwards=false
• 关键词: CAREER; Teaching; Machines; Design; Self

TECHNICAL SUMMARYThis CAREER award supports theoretical and computational research and education in the understanding and design of self-assembling biomaterials. Self-assembly of structured aggregates by the spontaneous organization of their constituent building blocks is prevalent in the natural world, and is an attractive route to fabricate artificial materials with desirable properties that cannot be easily produced by other means. The design of building blocks programmed to self-assemble custom materials is a grand challenge in materials science.In this work, the PI will integrate statistical mechanics theory with nonlinear machine learning algorithms to establish a new theoretical and computational approach to understand and program the self-assembly of nanostructured biomaterials. Using these tools, the PI will extract from molecular simulations the pathways and mechanisms by which building blocks self-assemble into structured aggregates. This methodology overcomes a key scientific challenge by integrating thermodynamics and kinetics in a unified framework that identifies both what stable aggregates form (thermodynamics) and how they assemble (kinetics and mechanisms). The collective order parameters unveiled by this approach are good descriptors of the slow dynamical motions driving assembly, and present a natural parameterization for kinetically meaningful free energy landscapes that link building block properties to collective assembly behavior. By "sculpting" the landscape topography through rational manipulation of building block structure and chemistry the PI's group will program the assembly of desired structures that are thermodynamically stable and kinetically accessible (design).The PI will apply a new approach to three technologically important self-assembling biomaterials: 1) "patchy colloid" polyhedral clusters for small molecule encapsulation, 2) ultra-short peptide mineralization templates for silica nanotubes for controlled drug release, heavy metal ion adsorption, and catalysis, and 3) antimicrobial peptide amphiphile nanostructures for antibiotic resistant bacteria. This work will establish new basic understanding and control of materials assembly, and accelerate development of new structural and functional biomaterials. The integrated education and outreach plan incorporates the scientific outcomes into education and outreach, and supports graduate training, undergraduate research, and mentoring of underrepresented minority groups. The PI will create a new materials science course to equip the next generation workforce with computational tools, support undergraduate students in performing portions of the work, and promote the recruitment, retention, and success of students of color through mentorship of minority students and high school outreach.NONTECHNICAL SUMMARYThis CAREER award supports a theoretical and computational research program to design microscopic building blocks with the ability to spontaneously self-organize into materials with desirable properties. This way of making materials is known as "bottom-up self-assembly", as opposed to more familiar "top-down" manufacturing. Imagine if it will be possible one day to design molecules with just the right shape and properties so that shaking them in a flask spontaneously self-assembled a solar cell! In this work, the PI will combine ideas from the fields of thermodynamics and machine learning (sometimes known as artificial intelligence) to establish a new tool to allow computers to learn both what structures can be formed by a particular building block, and how they assemble. The PI will then flip this problem to use our tool to help reverse-engineer building blocks to assemble custom materials. The PI's group will apply these tools to the design of three useful biological materials: 1) micron-sized particles possessing directional sticky patches that assemble polyhedral clusters to hold and deliver small molecules, 2) short peptides that assemble networks to template the synthesis of silica nanotubes for drug delivery, cleanup of heavy metal pollutants, and catalysis of chemical reactions, and 3) longer peptides that assemble into nanometer sized rods that can kill antibiotic resistant bacteria such as the MRSA "superbug".This award also supports an integrated research and education program in which the scientific results from this work will enrich and enhance undergraduate and graduate classes, and high school outreach activities. Undergraduate students will directly participate in the scientific research by working with the PI during the summer months. The PI will also design and teach a new class providing hands-on experience in the computational materials modeling, analysis, and design, and maintain his commitment to promote the recruitment and success of students of color through mentorship of undergraduate and graduate minority students.

该职业奖支持在理解和设计自组装生物材料方面的理论和计算研究和教育。结构化聚集体通过其组成构件的自发组织的自组装在自然界中是普遍的，并且是制造具有通过其他手段不能容易地产生的期望性质的人造材料的有吸引力的途径。设计可自组装定制材料的构建模块是材料科学的一项重大挑战。在这项工作中，PI将结合统计力学理论和非线性机器学习算法，建立一种新的理论和计算方法来理解和编程纳米结构生物材料的自组装。使用这些工具，PI将从分子模拟中提取构建块自组装成结构化聚集体的途径和机制。这种方法克服了一个关键的科学挑战，通过整合热力学和动力学在一个统一的框架，确定什么稳定的聚集体形式（热力学）和他们如何组装（动力学和机制）。这种方法所揭示的集体顺序参数是很好的描述缓慢的动力学运动驱动组件，并提出了一个自然的参数化动力学意义的自由能景观，链接积木属性的集体组装行为。通过合理操作积木结构和化学物质来“雕刻”景观地形，PI的团队将对所需结构的组装进行编程，这些结构具有化学稳定性和动力学可达性（设计）。PI将采用一种新方法来研究三种技术上重要的自组装生物材料：1）用于小分子包封的“片状胶体”多面体簇，2）用于控制药物释放、重金属离子吸附和催化的二氧化硅纳米管的超短肽矿化模板，和3）用于抗生素抗性细菌的抗微生物肽两亲物纳米结构。这一工作将建立对材料组装的新的基本认识和控制，并加速新型结构和功能生物材料的开发。综合教育和推广计划将科学成果纳入教育和推广，并支持研究生培训，本科生研究和指导代表性不足的少数群体。PI将创建一个新的材料科学课程，为下一代劳动力配备计算工具，支持本科生执行部分工作，并促进招聘，保留，通过少数民族学生的辅导和高中外展，有色人种学生的成功。非技术性总结这个职业生涯奖支持理论和计算研究计划，以设计微观积木的能力，自发地自组织成具有所需性质的材料。这种制造材料的方式被称为“自下而上的自组装”，而不是更熟悉的“自上而下”的制造。想象一下，如果有一天，我们能够设计出具有正确形状和性质的分子，那么在烧瓶中摇动它们，就可以自发地组装成太阳能电池！在这项工作中，PI将联合收割机从热力学和机器学习（有时称为人工智能）领域的想法，建立一个新的工具，让计算机学习什么结构可以形成一个特定的积木，以及他们如何组装。然后PI将翻转这个问题，使用我们的工具来帮助逆向工程构建块来组装自定义材料。PI的团队将应用这些工具设计三种有用的生物材料：1）具有定向粘性补丁的微米级颗粒，其组装多面体簇以保持和递送小分子，2）短肽，其组装网络以模板化用于药物递送、重金属污染物的清除和化学反应的催化的二氧化硅纳米管的合成，和3）更长的肽，组装成纳米大小的棒，可以杀死抗生素耐药细菌，如MRSA“超级细菌”。该奖项还支持一个综合的研究和教育计划，其中这项工作的科学成果将丰富和提高本科生和研究生课程，以及高中外展活动。本科生将直接参与科学研究，在夏季与PI一起工作。PI还将设计和教授一门新课程，提供计算材料建模，分析和设计方面的实践经验，并通过指导本科生和研究生少数民族学生，保持他促进有色人种学生招聘和成功的承诺。

项目成果

Landmark diffusion maps (L-dMaps): Accelerated manifold learning out-of-sample extension

• DOI: 10.1016/j.acha.2017.08.004
• 发表时间: 2019-07-01
• 期刊: APPLIED AND COMPUTATIONAL HARMONIC ANALYSIS
• 影响因子: 2.5
• 作者: Long, Andrew W.;Ferguson, Andrew L.
• 通讯作者: Ferguson, Andrew L.