Stacks Image 425
The principal goal of the Ferguson Lab is to establish detailed understanding of the equilibrium and dynamic behavior of soft materials through molecular simulation and machine learning. Our motivations are the design of next generation biomaterials and the development of molecular level insights to guide the development of therapeutics for infectious disease.
I. Folding and Self-Assembly of Biological and Bioinspired Materials. Self-assembly of complex aggregates is the driving force for the synthesis of complex cellular structures such as lipid membranes and viral capsids. Modern advances in the fabrication of micro and nanoscale “building blocks” have rendered self-assembly a potential synthesis route for novel materials such as 3D photonic crystals and drug delivery vessels, but the fundamental mechanisms and rules for robust synthesis of a desired structure remain poorly understood. We have developed a novel adaptation of a nonlinear machine learning technique rendering it extensible to the automated identification of self-assembly pathways from molecular simulations. As part of this research program, we are interested in the following goals:

  • Development of the collective diffusion map
  • Data mining of viral capsid assembly pathways
  • Determination of patchy colloid design rules
  • Simulation and data mining of the assembly paths of antimicrobial peptide nanostructures
  • “Inverse design” of tailored nanomaterials

Graduate Researchers: Andy Long, Rachael Mansbach, Jiang Wang
Undergraduate Researchers: Bridgette Lafaye, Suraj Dhanak

  • A.L. Ferguson, P.G. Debenedetti and A.Z. Panagiotopoulos “Solubility and molecular conformations of n-alkane chains in water” J. Phys. Chem. B 113 18 6405-6414 (2009) []
  • A.L. Ferguson, A.Z. Panagiotopoulos, P.G. Debenedetti and I.G. Kevrekidis “Systematic determination of order parameters for chain dynamics using diffusion maps” Proc. Natl. Acad. Sci. USA 107 31 13597-13602 (2010) []
  • A.L. Ferguson, S. Zhang, I. Dikiy, A.Z. Panagiotopoulos, P.G. Debenedetti and A.J. Link “An experimental and computational investigation of lasso formation in microcin J25” Biophys. J. 99 9 3056-3065 (2010) []
  • A.L. Ferguson, A.Z. Panagiotopoulos, I.G. Kevrekidis and P.G. Debenedetti “Nonlinear dimensionality reduction in molecular simulation: The diffusion map approach” Chem. Phys. Lett. Frontiers 509 1 1-11 (2011) []
  • A.W. Long and A.L. Ferguson* “Nonlinear machine learning of patchy colloid self-assembly mechanisms and pathways” J Phys. Chem. B 118 15 4228-4244 (2014) []
  • A.W. Long, J. Zhang, S. Granick, and A.L. Ferguson* "Machine learning assembly landscapes from particle tracking data" Soft Matter 11 8141-8153 (2015) []
  • R.A. Mansbach and A.L. Ferguson* "Machine learning of single molecule free energy surfaces and the impact of chemistry and environment upon structure and dynamics" J. Chem. Phys. 142 105101 (2015) []
  • A.W. Long, C.L. Phillips, E. Jankowski, and A.L. Ferguson* "Nonlinear machine learning and design of reconfigurable digital colloids" Soft Matter 12 7119-7135 (2016) []
  • A.W. Long and A.L. Ferguson* "Landmark diffusion maps (L-dMaps): Accelerated manifold learning out-of-sample extension" Appl. Comput. Harmon. Anal. (in press, 2017) []
  • R.A. Mansbach and A.L. Ferguson* "Control of the hierarchical assembly of π-conjugated optoelectronic peptides by pH and flow" Org. Biomol. Chem. 15 26 5484-5502 (2017) []
  • W.F. Reinhart, A.W. Long, M.P. Howard, A.L. Ferguson, and A.Z. Panagiotopoulos "Machine learning for autonomous crystal structure identification" Soft Matter 13 4733-4745 (2017) [pdf] []
  • J. Wang and A.L. Ferguson "A study of the morphology, dynamics, and folding pathways of ring polymers with supramolecular topological constraints using molecular simulation and nonlinear manifold learning" Macromolecules (submitted, 2017)
Stacks Image 434
Stacks Image 436
II. Accelerated Sampling in Biophysical Simulation. A generic problem in molecular simulation is the presence of high free energy barriers that prevent the simulation trajectory from adequately exploring the thermally accessible phase space. Biasing techniques seek to enhance sampling by driving simulations in a small number of order parameters to artificially enhance barrier crossing. A persistent difficulty in such approaches, however, is the availability of “good” biasing variables – typically associated with the important dynamical modes – in which to perform sampling. We have reformulated linear and nonlinear dimensionality reduction techniques to recover unbiased order parameters from biased simulation trajectories, facilitating an iterative algorithm to simultaneously recover good biasing variables and efficiently accelerate phase space exploration. We are deploying this powerful methodology in pursuit of the following goals:

  • Development of theory and implementation of autoencoder variable discovery and acceleration protocol
  • Initial validation of the approach for solvated polyalanine
  • Extension to realistic proteins (e.g., Trp cage) and those of therapeutic import (e.g. HIV env)
  • Plugin development to integrate the approach with simulation packages (e.g., PLUMED)

Graduate Researchers: Wei Chen
Undergraduate Researchers: Aik Rui Tan

  • A.L. Ferguson*, A.Z. Panagiotopoulos, P.G. Debenedetti and I.G. Kevrekidis “Integrating diffusion maps with umbrella sampling: Application to alanine dipeptide” J. Chem. Phys. 134 135103 (2011) []
  • A.L. Ferguson, A.Z. Panagiotopoulos, I.G. Kevrekidis and P.G. Debenedetti “Nonlinear dimensionality reduction in molecular simulation: The diffusion map approach” Chem. Phys. Lett. Frontiers 509 1 1-11 (2011) []
  • A.L. Ferguson* "BayesWHAM: A Bayesian approach for free energy estimation, reweighting, and uncertainty quantification in the weighted histogram analysis method" J. Comput. Chem. 38 18 1583-1605 (2017) []
  • J. Wang and A.L. Ferguson* "Nonlinear machine learning in simulations of soft and biological materials" Mol. Sim. (submitted, 2017)
Stacks Image 446
Stacks Image 3443
III. In Silico Viral Fitness Landscapes. Our third research focus is centered on the application and development of statistical mechanical tools to study viral fitness landscapes and evolutionary dynamics. Maximum entropy models parameterized from viral sequence databases permit the synthesis of effective Hamiltonians quantifying viral fitness in multidimensional sequence space. Such models for HIV have exhibited remarkable agreement with experimental fitness assays, escape mutations and compensatory mutational patterns, and have allowed us to quantitatively design Pareto optimal vaccine candidates. We are extending this modeling paradigm to other viral systems (influenza, dengue fever, hepatitis C), and are interested the following projects:

  • Development of Potts fitness landscapes
  • Application to deep sequencing data
  • Viral “phase transitions” in fitness space
  • Coupling of landscapes to evolutionary predator-prey dynamics
  • Machine learning of viral escape pathways
  • "Sloppiness" in viral Hamiltonians
  • Metastable basins in sequence space and viral phylogeny

Graduate Researchers: Greg Hart
Undergraduate Researchers: Sam Kaufman, Chin-Yu "Chester" Chen (ECE)

  • K. Shekhar, C.F. Ruberman, A.L. Ferguson, J.P. Barton, M. Kardar, A.K. Chakraborty "Spin models inferred from patient-derived viral sequence data faithfully describe HIV fitness landscapes" Phys. Rev. E 88 062705 (2013) []
  • A.L. Ferguson, E. Falkowska, L.M. Walker, M.S. Seaman, D.R. Burton and A.K. Chakraborty “Computational prediction of broadly neutralizing HIV-1 antibody epitopes from neutralization activity data” PLOS ONE 8 12 e80562 (2013) []
  • A.L. Ferguson, J.K. Mann, S. Omarjee, T. Ndung’u, B.D. Walker and A.K. Chakraborty “Translating HIV sequences into quantitative fitness landscapes predicts viral vulnerabilities for rational immunogen design” Immunity 38 606-617 (2013) []
  • J.K. Mann, J.P. Barton, A.L. Ferguson, S. Omarjee, B.D. Walker, A.K. Chakraborty and T. Ndung’u "The fitness landscape of HIV-1 gag: Advanced modeling approaches and validation of model predictions by in vitro testing" PLOS Comput. Biol. 10 8 e1003776 (2014) []
  • G.R. Hart and A.L. Ferguson* "Error catastrophe and phase transition in the empirical fitness landscape of HIV" Phys. Rev. E 91 032705 (2015) []
  • G.R. Hart and A.L. Ferguson* "Empirical fitness models for hepatitis C virus immunogen design" Physical Biology 12 066006 (2015) []
  • G.R. Hart and A.L. Ferguson "Viral fitness landscapes: A physical sciences perspective" in "Systems Immunology: An introduction to modeling methods for scientists" J. Das, C. Jayaprakash (eds.) Taylor and Francis (in press, 2017)
Stacks Image 459
Stacks Image 461
IV. Self Assembling Organic Electronics. In collaboration with the experimental labs of J.D. Tovar (Chem, Johns-Hopkins), Howard Katz (MSE, Johns-Hopkins), Bill Wilson (MRL, UIUC), JJ Cheng (MatSE, UIUC) and Charles Schroeder (ChBE, UIUC), we are investigating self-assembling organic electronics for energy transport at biotic/abiotic interface. Synthetic monomers comprising functionalized aromatic linkers flanked by short peptide sequences have demonstrated the capacity to self-assemble into high aspect ratio ribbons. Stacking of the aromatic linkers along the ribbon backbone leads to delocalization of the pi orbitals enabling conduction of electrons along the ribbon, presenting a biocompatible “wire” for bioelectronic applications. Using biomolecular simulations, we are probing the thermodynamics and kinetics of ribbon assembly at various resolutions, exploring the hierarchical structure of these ribbons, and developing rules for the rational engineering of more stable and robust bioorganic conductive ribbons. We are pursuing the following projects in this theme:

  • Simulation of the potential of mean force between peptide monomers in explicit solvent
  • Implicit solvent modeling of atomistically detailed peptide monomers
  • Simulations of assembly and evolution at long length and time scales with coarse grained models
  • Development of rational design rules to control backbone morphology with peptide sequence

Graduate Researchers: Bryce Thurston, Rachael Mansbach
Undergraduate Researchers: Deepak Mani

  • B.D. Wall, A.E. Zacca, A.M. Sanders, W.L. Wilson, A.L. Ferguson and J.D. Tovar "Supramolecular polymorphism: Tunable electronic interactions within pi-conjugated peptide nanostructures dictated by primary amino acid sequence" Langmuir 30 20 5946-5956 (2014) []
  • B.D. Wall, Y. Zhou, S. Mei, H.A.M. Ardona, A.L. Ferguson and J.D. Tovar "Variation of formal hydrogen bonding networks within electronically delocalized pi-conjugated oligopeptide nanostructures" Langmuir 30 38 11375-11385 (2014) []
  • B.A. Thurston, J.D. Tovar, and A.L. Ferguson* "Thermodynamics, morphology, and kinetics of early-stage self-assembly of pi-conjugated oligopeptides" Mol. Sim. 42 12 955-975 (2016) []
  • R.A. Mansbach and A.L. Ferguson* "Coarse-grained molecular simulation of the hierarchical self-assembly of π-conjugated optoelectronic peptides" J. Phys. Chem. B 121 7 1684-1706 (2017) []
  • R.A. Mansbach and A.L. Ferguson* "Control of the hierarchical assembly of π-conjugated optoelectronic peptides by pH and flow" Org. Biomol. Chem. 15 26 5484-5502 (2017) []
Stacks Image 495
Stacks Image 500
V. Lasso peptide structure, dynamics and stability. In collaboration with the experimental lab of Jamie Link (CBE, Princeton) we are studying the structure, stability, and maturation of so-called “lasso” peptides. Comprising approximately 20 amino acid residues, these proteins undergo post-translational modification to install an isopeptide bond between the N-terminal amino group and an acidic side chain that “captures” the C-terminal tail into a threaded lasso configuration. Possessing antimicrobial activity, structural rigidity, thermal stability, and denaturant resistance, the lasso motif is of extreme interest in protein engineering. We are conducting molecular modeling to understand the structure, dynamics, maturation, and thermal stability of these peptides, and to engineer tailored thermostabilty and functionality. We are pursuing the following projects:

  • Biased sampling exploration of the folding landscape of the capistruin linear precursor
  • Atomic structure, dynamics, and thermal stability of astexin-2 and -3
  • Evaluation of the reversible work of unthreading the astexin-3 lasso as a function of sequence

Graduate Researchers: Caitlin Allen (CBE, Princeton)
Undergraduate Researchers: Alex Trick, Maria Chen (CBE, Princeton)

  • A.L. Ferguson, S. Zhang, I. Dikiy, A.Z. Panagiotopoulos, P.G. Debenedetti and A.J. Link “An experimental and computational investigation of lasso formation in microcin J25” Biophys. J. 99 9 3056-3065 (2010) []
  • C.D. Allen, M.Y. Chen, A.Y. Trick, D. Thanh Le, A.L. Ferguson*, and A.J. Link "Thermal unthreading of the lasso peptides astexin-2 and astexin-3" ACS Chem. Biol. 11 11 3043-3051 (2016) []
Stacks Image 534
Stacks Image 537
VI. Inference of biomolecular folding landscapes from experimental time series. A beautiful result from dynamical systems theory is that the topography of the multidimensional free energy landscape governing the evolution of a high-dimensional system may be ascertained by appropriate processing of time series in a single generic system variable. By combing these techniques with advanced nonlinear manifold learning algorithms, we have developed a means to infer protein folding landscapes from univariate time series recordings of, for example, the protein radius of gyration. We are developing and testing these new tools in biomolecular simulations of protein folding to establish conditions on the time series and acceptable levels of sampling noise to enable their application to experimental single molecule FRET data. We are pursuing the following projects in this theme:

  • Comparison of landscapes from full-dimensional and univariate simulation trajectories
  • Inference and conditions on the Jacobian of the diffeomorphism linking the two landscapes
  • Investigation of the effects of sampling noise and uncertainty on the landscape
  • Modeling of system evolution over these landscapes using low-dimensional Langevin equations
  • Partnering with single molecule biophysicists to recover experimental folding landscapes

Graduate Researchers: Jiang Wang, Matt Ellis (Math)

Stacks Image 547
VII. Mesoscale modeling and simulation of asphaltene aggregation. Crude oil is comprised of three main composition groups: paraffinic, napthenic and aromatic. In the aromatic group, benzene is the lightest component and asphaltenes the heaviest. Asphaltenes comprise the heaviest aromatic fraction and possess a propensity to aggregate and precipitate out of solution during petroleum processing. Aggregation is thought to proceed according to the Yen-Mullins hierarchy, but the molecular mechanisms underlying mesoscopic assembly remain poorly understood. We are developing coarse-grained molecular models explicitly parameterized against all-atom calculations to conduct simulations capable of attaining the time and length scales necessary to directly simulate the complete Yen-Mullins hierarchy and ascertain the molecular mechanisms of assembly. We are also applying nonlinear machine learning to infer low-dimensional assembly landscapes possessing a continuum to discrete transition reminiscent of the random energy model in protein folding, and combining high-throughput simulation with supervised machine learning to rationally design of aggregation inhibitors. We are pursuing the following projects in this theme:

  • Development of coarse-grained mesoscale asphaltene models
  • Nonlinear learning of asphaltene aggregation pathways and mechanisms
  • High-throughput simulation and supervised machine learning for rational design of aggregation inhibitors
  • Development of design rules for aggregation behavior as a function of asphaltene chemistry and structure

Graduate Researchers: Jiang Wang
Undergraduate Researchers: Mohit Gayatri

Stacks Image 1434
VIII. Molecular simulation and machine learning discovery and design of antimicrobial membrane active peptides. Antimicrobial peptides (AMPs) comprise a critical component of innate immunity in plants and animals. There are some ~1100 known antimicrobial peptides that possess broad spectrum antimicrobial activity by non-specific antimicrobial interactions with bacterial membranes. Although AMPs tend to be short, cationic, and amphiphilic, their enormous sequence diversity defies the definition of simple design rules. Moreover, there is evidence for manifold mechanisms of activity beyond simple membrane permeabilization, including inhibition of nucleic acid and protein synthesis and disruption of enzyme activity. In a collaboration with experimentalists Gerard Wong (BioE, UCLA) and Jianjun Cheng (MatSE, UIUC) we are employing molecular dynamics simulations and supervised machine learning to navigate structure and sequence space to discover physicochemical commonalities defining the "blueprint" for antimicrobial activity. We are validating and improving our models in a closed loop and self-reinforcing cycle between computation and experiment, and using our approach to identify new membrane active peptides with divergent sequences from any known natural AMPs, discovering membrane activity in broad families of sequences previously unknown to be membrane active, and engineering multiplexed functionality into synthetic membrane active peptides. We are pursuing the following goals in this theme:

  • Molecular engineering of radial amphiphilicity
  • Development of supervised machine learning tools to navigate AMP sequence space and discover AMP design rules
  • Molecular simulation of triggerable / switchable amphiphilic AMPs
  • Discovery and engineering of multiplexed functionality into membrane active peptides

Graduate Researchers: Ben Fulan (Math), Ernest Lee (BioE, UCLA)

  • M. Xiong, M.W. Lee, R. Mansbach, Z. Song, Y. Bao, R.M. Peek Jr., C. Yao, L.-F. Chen, A.L. Ferguson*, G.C.L. Wong, and J. Cheng "Helical antimicrobial polypeptides with radial amphiphilicity" Proc. Natl. Acad. Sci. USA 112 43 13155-13160 (2015) []
  • E.Y. Lee, B.M. Fulan, G.C.L. Wong, and A.L. Ferguson* "Mapping membrane activity in undiscovered peptide sequence space using machine learning" Proc. Natl. Acad. Sci. USA 113 48 13588-13593 (2016) []
  • E.Y. Lee, M.W. Lee, B.M. Fulan, A.L. Ferguson*, and G.C.L. Wong "What can machine learning do for antimicrobial peptides, and what can antimicrobial peptides do for machine learning?" Interface Focus (in press, 2017)
  • Z. Song, R.A. Mansbach, R. Baumgartner, K.-C. Shih, H. He, N. Zheng, X. Ba, Y. Huang, D. Mani, Y. Lin, M.-P. Nieh, A.L. Ferguson*, L. Yin, and J. Cheng "Modulation of polypeptide conformation through donor-acceptor transformation of side-chain hydrogen bonding ligands" Nat. Commun. 92 8 1-8 (2017) []
  • E.Y. Lee, G.C.L. Wong, and A.L. Ferguson* "Machine learning discovery and design of membrane-active peptides" Bioorg. Med. Chem. (in press, 2017) []
  • M.W. Lee, E.Y. Lee, G.H. Lai, N.W. Kennedy, A.E. Posey, W. Xian, A.L. Ferguson, R.B. Hill, and G.C.L. Wong "Molecular motor Dnm1 synergistically induces membrane curvature to facilitate mitochondrial fission" ACS Cent. Sci. (submitted, 2017)
Stacks Image 3448
Stacks Image 3450
IX. Undergraduate Education. There is consensus among engineering departments in US institutes of higher education, driven by discussions with government, academic, and national lab stakeholders, that undergraduates require a deeper understanding of computational tools. The 2011 White House Materials Genome Initiative to accelerate the development of new materials asserts that computer-aided materials design and training of the next generation workforce in these tools is vital to national competitiveness and welfare. Skills in computational modeling and design are desirable to employers hiring engineering graduates into industrial R&D and product development positions, national labs, and academic research positions. These trends mandate that engineering departments in universities around the country must better serve their graduates, industry, and the nation by better equipping students with training in computational tools. As part of a team of six faculty in the Department of Materials Science and Engineering we are integrating computational modules into the core undergraduate curriculum (MSE 201 – Phases and Phase Relations, MSE 206 – Mechanics for MatSE, MSE 304 – Electronic Properties of Materials, MSE 406 – Thermal and Mechanical Behavior of Materials, MSE 404 – Computational MatSE) in order to provide training in computational materials science tools and improve learning and engagement with the course material. As part of these education innovations we are pursuing the following goals:

  • Provide undergraduates with training in academic and industrially relevant computational materials science tools
  • Develop and integrate computational modules into undergraduate MatSE curriculum
  • Improve learning and active engagement with course concepts and material through computation
  • Establish a computational concept inventory and enhance computational thinking

Graduate Researchers: Rachael Mansbach

  • R.A. Mansbach, G.L. Herman, M. West, D.R. Trinkle, A.L. Ferguson, and A. Schleife "WORK IN PROGRESS: Computational Modules for the MatSE Undergraduate Curriculum" Paper presented at American Society for Engineering Education (ASEE) 123rd Annual Conference & Exposition, New Orleans, LA, June 26-29 2016 []
  • A. Kononov, P. Bellon, T. Bretl, A.L. Ferguson, G.L. Herman, K.A. Kilian, J.A. Krogstad, C. Leal, C.R. Maass, A. Schleife, J.K. Shang, D.R. Trinkle, and M. West "Computational Curriculum for MatSE Undergraduates" Paper presented at 2017 American Society for Engineering Education (ASEE) 124th Annual Conference & Exposition, Columbus, OH, June 25-28 2017 []
  • R.A. Mansbach, A.L. Ferguson, K.A. Kilian, J.A. Krogstad, C. Leal, A. Schleife, D.R. Trinkle, M. West, and G.L. Herman "Reforming an undergraduate materials science curriculum with computational modules" J. Mater. Educ. 38 3-4 161-174 (2016)
Stacks Image 2383
X. Other. We have a number of miscellaneous interests that defy categorization into our primary research thrusts including:

  • nanoscopic hydrophobicity
  • nonlinear dimensionality reduction methods development
  • Bayesian reformulations of free energy landscape construction
  • graph matching algorithm development
  • anticancer nanoparticle modeling
  • peptide engineering

Graduate Researchers: Andy Long, Rachael Mansbach
Undergraduate Researchers: Jingtian Hu, Ali Hajimirza, Abhijit Pujare

  • A.L. Ferguson*, N. Giovambattista, P.J. Rossky, A.Z. Panagiotopoulos and P.G. Debenedetti “A computational investigation of the phase behavior and capillary sublimation of water confined between nanoscale hydrophobic plates” J. Chem. Phys. 137 144501 (2012) []
  • L. Tang, X. Yang, I. Chaudhury, C. Yao, Q. Yin, Q. Zhou, M. Kwon, L.W. Dobrucki, L.B. Borst, S. Lezmi, W.G. Helferich, A.L. Ferguson*, T.M. Fan and J. Cheng "Investigating the optimal size of anticancer nanomedicine" Proc. Natl. Acad. Sci. USA 111 43 15344-15349 (2014) []
  • J. Hu and A.L. Ferguson* "Global graph matching using diffusion maps" Intelligent Data Analysis 20 3 637-654 (2016) []
Stacks Image 2391