Bulletin of the American Physical Society
APS March Meeting 2023
Volume 68, Number 3
Las Vegas, Nevada (March 5-10)
Virtual (March 20-22); Time Zone: Pacific Time
Session LL10: V: Data Science I |
Hide Abstracts |
Sponsoring Units: GDS Chair: William Ratcliff, National Institute of Standards and Technology; University of Maryland Room: Virtual Room 10 |
Tuesday, March 21, 2023 5:00AM - 5:12AM |
LL10.00001: Design of β-sheet forming antimicrobial peptides using deep learning and molecular dynamics simulations Rachael A Mansbach, Samuel Renaud, Lindsay Wright, Mohammadreza Niknam Hamidabad, Natalya Watson Antimicrobial peptides (AMPs) represent a potent alternative to traditional antibiotics treatments. They are typically short peptides created by the innate immune system that preferentially attach themselves to the cell walls of bacteria and lyse them, through a number of different biophysical mechanisms. Design of such peptides can be slow due to the length of time it takes to assess their properties, whether computationally or experimentally. Machine learning methods have recently emerged as powerful tools for efficient design of biomolecules. In this talk, I will discuss our efforts to instantiate a computational workflow for active learning design of AMPs through search space design and molecular dynamics assessment of points in the search space. To investigate deep learning for instantiation of continuous search spaces, we created different AMP search spaces produced by different deep learning model architectures. We assessed reconstruction accuracy, generative capability, and model interpretability and demonstrate that while most models are able to create a partitioning in their latent spaces into regions of low and high AMP sampling probability, they do so in different manners. In this way we demonstrated several benchmarks that must be considered for such models and suggest that for optimization of search space properties, an ensemble methodology is most appropriate for design of new AMPs. To assess qualities of points in the latent space, we focused on a synthetic beta-sheet forming antimicrobial peptide, GL13K, and performed molecular dynamics simulations of it and several variants. We show that in the presence of negatively-charged bacterial membranes, it undergoes aggregation followed by conformational rearrangement into larger disordered clumps of smaller clusters. Overall, our work provides an important step towards rational and efficient design of AMPs to combat growing antimicrobial resistance. |
Tuesday, March 21, 2023 5:12AM - 5:24AM |
LL10.00002: Improved Knowledge Representation Enables AI-Guided Polymer Design and Experimental Validation Nathan Park The delivery of actionable predictions or hypothesis is a key outcome of any predictive modeling effort for experimental science. Within the context of polymer chemistry—where experimentalists must navigate a highly complex design space—effective knowledge representation of experimental becomes critical to create useful models with actionable predictions. Here, we detail our efforts on the development of new, extensible open-source tools to enable experimentalists to accurately represent experimental data and facilitate its consumption in AI/ML or informatics pipelines. Moreover, we will demonstrate how these tools enabled the development and validation of generative models for new polymerization catalysts and how these generated catalysts can be repurposed in adjacent application areas within small-molecule organic chemistry. Finally, the integration of these toolkits within a growing ecosystem of open-source platforms polymer AI/ML and polymer data repositories will be discussed. |
Tuesday, March 21, 2023 5:24AM - 5:36AM |
LL10.00003: The aerodynamic characteristics are modeled based on machine learning Xing Zhao Aerodynamic characteristic modeling mainly includes mechanism modeling method and "black box" modeling method. This paper analyzes and applies the machine learning methods of "black box" modeling - Classification and regression tree method and shallow learning method. The classification and regression tree method, Kriging modeling method, RBF neural network method and SVM support vector machine method in shallow learning method are applied to rocket aerodynamic characteristic modeling, delta wing unsteady aerodynamic characteristic modeling at high angle of attack and aerodynamic thermal test data fusion respectively. The advantages and disadvantages of these modeling methods are compared and analyzed, and better prediction results are obtained, The application scope of machine learning modeling method for aerodynamic characteristics is expanded. |
Tuesday, March 21, 2023 5:36AM - 5:48AM |
LL10.00004: Flagging of unacceptable segmentations: Monte Carlo dropout vs. Deep-Ensembles Zan Klanecek, Tobias Wagner, Yao K Wang, Lesley Cockmartin, Nicholas Marshall, Brayden Schott, Ali Deatsch, Miloš Vrhovec, Andrej Studen, Hilde Bosmans, Robert Jeraj Knowing when your deep learning model is producing inadequate segmentations is crucial. In this work, we leveraged the quantification of predictive uncertainty (PU) to flag unacceptable pectoral muscle segmentations in mammograms. Two methods were compared for the estimation of PU: Monte Carlo (MC) dropout and Deep-Ensembles (DE). |
Tuesday, March 21, 2023 5:48AM - 6:00AM |
LL10.00005: Numerical generation and coating of single soot nanoparticles: implications for atmospheric aging Cyprien Jourdain, Adam Boies, Jonathan Symonds The adverse effects due to the release of black carbon-based nanoparticles in the atmosphere represent the second anthropological cause of global warming and constitute one of the largest uncertainties in climate modeling. The lifecycle of such particles is complex: the freshly emitted soot ages over a timescale of minutes to days via condensation and coagulation of weakly absorbing organic aerosols. The transport, interaction with clouds, and the induced radiative forcing directly depend on the morphological and optical properties of these coated nanoparticles. This work proposes a standalone model that generates bare and coated aggregates, and measures their morphological and transport-derived properties, as well as providing an interface for optical modeling. The model is applied to atmospheric aging of soot with water, sulfuric acid, or p-xylene coatings. Both partially and fully coated particles show enhanced absorption, scattering, and single scattering albedo at mid-visible wavelength. These values are maximum for fully encapsulated aggregates. A quasi linear relationship is found between the mass absorption cross-section fraction (fMAC) and the aerodynamic growth factor. Furthermore, fMAC is higher for aggregates composed of polydisperse primary particles. Other key parameters investigated include the coating material and its partitioning, the morphology of the aggregate (fractal parameters, size, overlap) and the incident wavelength. |
Tuesday, March 21, 2023 6:00AM - 6:12AM Author not Attending |
LL10.00006: Structural Investigation of Small Molecule Selectivity for Cardiolipin Bernadette Mohr, Tristan Bereau, Diego van der Mast Efficient development of new materials and drugs is becoming increasingly relevant. For targeted molecular design, information is needed about the physical and chemical properties governing function. An automated approach for identifying the chemical interactions crucial for binding selectivity of a drug-like compound to a biomolecular target was recently demonstrated[1]. But the mere presence of these interactions is only part of the picture, in this work we investigate the structure-property relationship of a small molecule candidate with respect to target selectivity. Extracting mechanistic insights from MD simulations is a complex problem, with exact quantum-chemical solutions often computationally infeasible. Instead, we adapt the atomistic SLATM[2] representation for use on already present coarse-grained trajectory data to systematically obtain 1-, 2-, and 3-body interactions between small molecules and their environment. By correlating these interaction modes with their corresponding physical and chemical properties, we gain insight into the structural aspects of binding selectivity of candidate compounds to the phospholipid Cardiolipin. |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700