Bulletin of the American Physical Society
Mid-Atlantic Section Fall Meeting 2020
Volume 65, Number 20
Friday–Sunday, December 4–6, 2020; Virtual
Session G01: Machine Learning in Physics and Quantitative Biology |
Hide Abstracts |
Chair: Justin Kinney |
Saturday, December 5, 2020 4:15PM - 4:51PM |
G01.00001: Massively Parallel Reporter Assays, Machine Learning, and the Biophysics of Gene Regulation Invited Speaker: Justin Kinney Gene expression in all organisms is controlled by short DNA and RNA sequences called cis-regulatory elements (CREs). Proteins in the cellular milieu bind to nucleic acid sequences present within CREs, interact with one another, and thus form macromolecular complexes that modulate the expression of nearby genes. My lab uses a combination of experiments and mathematical modeling to study the biophysics of these regulatory processes in living cells. Our experiments are based on massively parallel reporter assays, which leverage the power of ultra-high-throughput DNA sequencing to measure the effects that thousands to millions of different CRE sequence perturbations have on gene expression. Our mathematical modeling efforts aim to infer biophysical models from the large DNA sequence datasets that these experiments yield. This machine learning task has proven to be remarkably fertile from a theoretical standpoint, tying together ideas in Bayesian field theory, information theory, dimensionality reduction, and deep learning. [Preview Abstract] |
Saturday, December 5, 2020 4:51PM - 5:27PM |
G01.00002: Using Information Geometry to Find Simple Models of Complex Processes Invited Speaker: Mark Transtrum Effective theories play a fundamental role in how we reason about the world. Although real physical processes are very complicated, useful models abstract away the irrelevant degrees of freedom to give parsimonious representations. I use information geometry to construct simplified models for many types of complex systems, such as biology, neuroscience, statistical physics, and complex engineered systems. I interpret a multi-parameter model as a manifold embedded in the space of all possible data, with a metric induced by statistical distance. These manifolds are often bounded and very thin, so they are well-approximated by a low-dimensional, simple model. For many types of models, there is a hierarchy of natural approximations that reside on the manifold's boundary. These approximations are not black-boxes. They remain expressed in terms of the relevant combinations of mechanistic parameters and reflect the physical principles on which the complicated model was built. They can also be constructed in a systematic way using computational differential geometry. [Preview Abstract] |
Saturday, December 5, 2020 5:27PM - 5:39PM |
G01.00003: Predicting Physical Properties of Proteins Using 3D Convolutional Neural Networks. Talant Ruzmetov, Siddharth Bhadra-Lobo, Devlina Chakravarty, Guillaume Lamoureux Convolutional neural networks (CNNs) have gained widespread popularity by achieving state-of-the-art results on various image analysis tasks such as handwriting recognition, object detection, and medical diagnosis. An inherent property of CNNs is the so-called "translational equivariance", which enables the network to recognize an object (or pattern) irrespective of its position in the image. We use 3D CNNs as well as their rotationally equivariant versions, called SE(3)-CNNs, to predict spatially-resolved physical properties of proteins, such as the electrostatic potential, the density of solvent molecules, or the binding free energy of small molecular fragments. The binding free energies of small molecular fragments are obtained from molecular dynamics-based methods such as Site Identification by Ligand Competitive Saturation (SILCS). In this talk, we demonstrate how feature representations learned from protein atomic densities can be used to predict various physical properties of molecular interactions with 3D CNN-based architectures. [Preview Abstract] |
Saturday, December 5, 2020 5:39PM - 5:51PM |
G01.00004: Mathematical Models for Living Forms in Medical Physics Submodel 1: The Information Processing from Teeth to Nerves Christina Pospisil This talk continues the presentation at APS March Meeting 2019 and APS April Meeting 2019. In this part of the project the first submodel is presented. The information processing from teeth to the nerves. Information processing is modeled via p-waves passing through the tooth layers enamel and dentin. Odontoblasts located in the liquid in the tubules of the tooth dentin layer perform finally the transformation into electrical information (an electrical signal) that passes along nerves. The presentation was scheduled for the APS March Meeting 2020 Conference (the APS March Meeting 2020 Conference got canceled because of Covid-19), the presentation was given at the APS April Meeting 2020 Conference. [Preview Abstract] |
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