Bulletin of the American Physical Society
APS March Meeting 2022
Volume 67, Number 3
Monday–Friday, March 14–18, 2022; Chicago
Session A06: Data Science for BiophysicsFocus Recordings Available
|
Hide Abstracts |
Sponsoring Units: DBIO GDS DCOMP GSNP Chair: Steve Press Room: McCormick Place W-178B |
Monday, March 14, 2022 8:00AM - 8:36AM |
A06.00001: Bayesian nonparametrics in multi-particle superresolved tracking Invited Speaker: Zeliha Kilic Superresolved localization with experimental means such as PALM and STORM has previously been used to probe static molecular structures relevant to neurodegenerative diseases by resolving particles below the diffraction limit. However, superresolved localization does not provide any insight regarding the underlying dynamical processes. |
Monday, March 14, 2022 8:36AM - 8:48AM |
A06.00002: Inference of C. elegans neural network structure from calcium flurescence time series data with reservoir computing Amitava Banerjee, Sarthak Chandra, Edward Ott Calcium imaging is a popular technique to record neural activity from live, freely-moving animals, like C. elegans or mice. Inference of neural network connections from short duration, low temporal resolution time series data of calcium flurescence from individual neurons or neuron groups is important in understanding how neurons interact together to orchestrate behavior or stimulus response in animals. Previous works in this direction have mostly focused on inferring such neural interactions by measuring correlations and information flow between calcium flurescence time series of different neurons, without having an understanding of the underlying neural dynamics. In this work, we train a particular kind of machine learning architechture, called "reservoir computing", on publicly available whole-brain calcium imaging time series datasets from C. elegans. The trained reservoir is able to learn a model of the C. elegans neural dynamics, which we further use to estimate the connectivity structure between the neurons (this is based on our previous works, reported in [1], [2]). We validate our results against the known C. elegans neural connectome, and show that our neural connectivity inference method performs better than transfer entropy-based method, a commonly used technique for network inference from biological data. These results indicate that data-driven and machine learning-based modeling of neural dynamics has the potentiality to outperform traditional network inference techniques. |
Monday, March 14, 2022 8:48AM - 9:00AM |
A06.00003: Towards Interpretable Imputation of Missing Observations with Time Slice Synthetic Minority Oversampling Technique Andrew Baumgartner Dealing with sparse and irregular time series data presents numerous problems for comparing the trajectories of time dependent systems in a controlled manner. Particularly, data points which are sampled at different times do not allowed for such a controlled comparison since the behavior of two independent signals will behave differently at different times. Further, the irregularity of this type of data makes it difficult to use state-of-the-art machine learning techniques such as recurrent neural networks due to the rigidity of their architectures. To deal with these issues, we developed a simple yet novel non-parametric time series imputation technique with the goal of constructing an irregular time series that is uniform across every sample in a data set. Specifically, we fix a grid defined by the midpoints of non-overlapping bins (dubbed "slices") of observation times and ensure that each sample has values for all of the features at that given time. This allows one to both impute fully missing observations to allow uniform time series classification across the entire data and, in special cases, to impute individually missing features. We illustrate the technique in a number of examples and proof-of-concept directions for future research in biology and medicine. |
Monday, March 14, 2022 9:00AM - 9:12AM |
A06.00004: Graph neural networks for multicellular dynamics Ming Han, John Devany, Margaret Gardel, Vincenzo Vitelli Cells in a tissue often form a disordered network. It mediates complex intercellular interactions, which strongly affect the migration and division of the tissue cells. Such multicellular dynamics plays a crucial role in many biological processes ranging from wound healing to organogenesis. However, given the complexity of the internal drive of individual cells and their interactions, it is extremely difficult to establish a theoretical model from first principle. Here we propose a generic machine learning approach capable of learning various multicellular dynamics from recorded experimental videos. Instead of requiring the internal states of living cells that are hard to access, our model relies solely on the external geometric information of them (i.e. cell shape, size, and interconnectivity) that are easy to measure. To learn the cell interactions that can be both non-reciprocal and pathway-dependent, we represent a tissue system in terms of both cell and interaction graphs and apply advanced neural networks onto this dual graph representation. Taking epithelial cells as an illustrative example, we show our graph neural network not only captures the stochastic cell motion but also predicts the evolution of cell states in their division cycle. We show this method can be easily extended to forecast the developmental systems (e.g. fly wing cells) and the cell signaling process (e.g. ERK signaling). |
Monday, March 14, 2022 9:12AM - 9:24AM |
A06.00005: Sociophysics of subjective pain: robust cluster assignment based on sparsely and irregularly sampled data from a dynamical system Gary K Nave, Swati Padhee, Kumar Utkarsh, Amanuel Alambo, Tanvi Banerjee, Nirmish Shah, Daniel M Abrams Prior research has attempted to explain changes in pain levels over time using differential equation based models. In this work, we use techniques drawn from the study of complex networks to motivate partitioning of patient populations into distinct clusters. We introduce and test a variety of methods for comparing irregularly and sparsely sampled trajectories, with the goal of providing a rigorous basis for defining clusters and assigning trajectories to them. |
Monday, March 14, 2022 9:24AM - 9:36AM |
A06.00006: Learning the shape of the protein universe with 3D-equivariant holographic convolutional neural networks Michael Pun, Andrew Ivanov, Quinn Bellamy, Colin LaMont, Jakub Otwinowski, Armita Nourmohammad Proteins are the machinery of life facilitating the key processes that drive living organisms. The physical arrangement of amino acids dictates how proteins fold and interact with their environment. Recent advances have increased the number of experimentally resolved or computationally predicted tertiary structures, however we still lack a practical understanding of how 3D structure determines the function of a protein. While machine learning has been at the forefront of protein science, the inferred models are often hard to interpret physically. Here, we introduce holographic convolutional neural networks (H-CNNs) that take atomic coordinates of a protein structure as input and, through 3D equivariant transformations that respect the rotational symmetries in data, learn interpretable models of protein micro-environments reflecting the underlying biophysics. With H-CNNs, we infer amino acid preferences given a surrounding atomic neighborhood and predict the impact of evolutionary substitutions in proteins. Our computational approach establishes an interpretable model for how biological function emerges from protein micro-environments. The flexibility and efficiency of H-CNNs also show promise for building generative models to design novel protein structures with desired function. |
Monday, March 14, 2022 9:36AM - 9:48AM |
A06.00007: Machine learning continuum models for cellular force generation Matthew Schmitt, Jonathan Colen, Stefano Sala, Margaret Gardel, Patrick W Oakes, Vincenzo Vitelli Mechanical behaviors of cells arise through the mechanochemical interactions of proteins which self-organize into organelles and cytoskeletal structures. However, no systematic strategy exists to identify the relevant collective variables representing protein distributions within the cell and link these to mechanical response at the cellular scale. Here, we show how machine learning can be used to build continuum models that relate protein distributions to forces. We train neural networks to map between fluorescent protein distributions and experimental traction stresses and observe that focal adhesion proteins alone are sufficient for accurate force predictions. By calculating the importance the network assigns to features of these protein distributions, we identify relevant analytical terms in a gradient expansion of the input protein signal. After performing sparse regression on a neural network-inspired library of terms, we obtain continuum equations relating protein localization and cell stresses. |
Monday, March 14, 2022 9:48AM - 10:00AM |
A06.00008: EMBED: a low dimensional reconstruction of gut microbiome dynamics based on ecological normal modes Mayar A Shahin, Purushottam Dixit, Brian Ji
|
Monday, March 14, 2022 10:00AM - 10:12AM |
A06.00009: Parameter estimation from stochastic processes applied to neural dynamics and aging Helmut H Strey, Lilianne R Mujica-Parodi, Uri Alon, Yifan Yang H. H. Strey, Estimation of parameters from time traces originating from an Ornstein-Uhlenbeck process. Phys Rev E 100, 062142 (2019). |
Monday, March 14, 2022 10:12AM - 10:24AM |
A06.00010: Salient projections over maximum projections: Improving deep-learning detection and segmentation during invasion of cell-dense spheroids. Christopher Z Eddy Recently, cell shape analysis is boosted by techniques from computer vision. As a result, it becomes possible to obtain high content information of cellular states from morphological data alone. In this work, we present on a highly adaptable deep-learning module which accepts 3D inputs to propogate 3D spatial latent embeddings as inputs to 2D instance segmentation neural networks. We apply the modified networks to 3D time-lapse images of MDA-MB-231 cell-dense spheroids to show improved continuous detection and quality of segmentations of individual cells during invasion. We show that the segmentations made by the modified 2D networks are markedly improved in comparison to maximum projection input images, making vital improvements for confocal imaging experiments with low axial resolution. |
Monday, March 14, 2022 10:24AM - 10:36AM |
A06.00011: Non-linear interpolation for genetic fitness prediction Kai Shimagaki, John P Barton In numerical integration based on discrete data points, the choice of integration methods is crucial. Applying Itô or Stratonovich integration that is commonly used in stochastic calculus can lead to significantly different outcomes [1]. In this talk, we propose a stochastic non-linear interpolation method using Bézier’s method [2]. This method is simple and broadly applicable to physical systems. As an example, we apply our approach to the Marginal Path Likelihood method, a genomic fitness inference method based on the stochastic processes and formulated through a path integral [3]. |
Monday, March 14, 2022 10:36AM - 10:48AM |
A06.00012: Multi-Model Analysis of De novo Antimicrobial Peptide Design Via Variational Autoencoder Latent Sampling Samuel Renaud, Ré A Mansbach Growing concerns over antibiotic resistance have promulgated the development of novel therapeutic agents to treat microbial infections. Anti-microbial peptides (AMP) are short proteins that possess unique antimicrobial properties capable of overcoming antibiotic resistance through unique mechanisms of action. This work develops a comparison of generative deep learning models used widely in the field of chemical compound discovery as an application to de novo AMP generation. We analyze the competency of 5 architectures in generating novel anti-microbial peptides. The models investigated are adversarial autoencoders networks (AAE), variational auto-encoders (VAE), Wasserstein auto-encoders (WAE), VAE’s with an attention layer (AVAE) and VAE-Transformers (VAE-Trans). The VAE framework generates smooth explorable latent spaces for which interpretability mechanisms are presented. New AMP candidates with desirable features are sampled and verified from the continuous latent spaces using a feature prediction network. |
Monday, March 14, 2022 10:48AM - 11:00AM |
A06.00013: Learning the forces in active matter from the trajectories: a Graph-neural-network approach Miguel Ruiz-Garcia, Miguel Barriuso Gutierrez, Luca M Ghiringhelli, Chantal Valeriani Active particles exhibit complex collective phenomena that emerges from their local interactions. To model such systems, one would usually propose some inter-particle interactions and active forces, simulate the dynamics of a system with many individual elements and finally compare the results with experiments via, for instance, an order parameter. However, not only choosing one order parameter might introduce a bias, but also it is difficult to assess how well the model describes the experimental system. In our work we suggest a completely different approach. What if we could learn the inter-particle interactions and the active forces directly from the data? We propose a graph-neural-network-based scheme that learns the interactions between particles and the active forces to predict the correct particle dynamics. After training the network, one can extract both passive and active interactions between particles and use them (analytically or numerically) to make new predictions or unravel dynamical features of experiments of active particles. |
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