Carnegie Mellon University

Joint Carnegie Mellon-University of Pittsburgh Ph.D. Program in Computational Biology

Joint Carnegie Mellon-University of Pittsburgh Ph.D. Program in Computational Biology › Seminar Series › Speaker Abstracts

Date	Speaker	Abstract
September 16, 2022	Wayne Stallaert	Mapping cell cycle plasticity during tumorigenesis and treatment Understanding the organization of the cell cycle has been a longstanding goal in cell biology. In the past, we have used metaphors to describe this organization – a clock, a set of dominoes, an oscillator. But what if we could zoom out and actually see the cell cycle? Using a combination of hyperplexed, single-cell imaging and machine learning, we made a map of the human cell cycle. To make this map, we first performed iterative immunofluorescence to measure 50 cell cycle proteins in thousands of individual cells to obtain a high-dimensional signature of cell cycle state. We then used a nonlinear dimensionality reduction approach to place cells with similar cell cycle states close to each other in a lower dimensional embedding or “map”. These cell cycle maps resolved the fate trajectories of cells through both the proliferative cell cycle (G1/S/G2/M) and into cell cycle arrest (G0) and revealed heterogeneity in cell cycle progression among genetically identical cells. Importantly, these maps can also be used to measure how the cell cycle changes in response to different biological or disease contexts. Mapping the fate trajectories of cells following hypomitogenic, replicative and oxidative stress, for example, revealed that cell cycle arrest is not a discrete state but rather a complex and continuous architecture of cell states. Building upon this initial work, my lab will use cell cycle mapping to study the cancer cell cycle in its natural habitat: embedded within a complex, multicellular environment using patient tissues and organoid models. We will use this platform to investigate how the tumor microenvironment shapes the cancer cell cycle, the contribution of cell cycle plasticity to drug resistance and cell cycle arrest states that underlie cancer cell dormancy.
September 23, 2022	John Barton	Learning about viral evolution from temporal genetic data Pathogens such as HIV and SARS-CoV-2 pose serious threats to public health. Better understanding of how these pathogens evolve could inform efforts to control outbreaks and improve the design of vaccines. In this talk, I’ll discuss how we can model the evolution of viruses quantitatively. I’ll also show how we can use ideas from areas like statistical physics to fit quantitative evolutionary models to data, allowing us to understand how different mutations affect viral replication or transmission. We’ll explore two examples: understanding how HIV evolves within patients to escape from the immune system and how SARS-CoV-2 has evolved to become more transmissible over the course of the pandemic.
September 30, 2022	Wilson Wong	Mammalian Cell Design using Synthetic Biology In this talk, I will go over some of the work from our lab related CAR T cell technologies for cellular immunotherapy, such as a couple of logic CAR platforms, and drug-gated CAR circuits. I will also discuss our development in gene expression control technologies, especially related to recombinases and CRISPR Cas13.
December 2, 2022	Amy Goldberg	Evolutionary Perspectives on Malaria Classically called one of the strongest selective pressures in human evolution, I will discuss the ongoing host and pathogen pressures shaping malaria disease. First, considering the added information that distributions of genetic ancestry provide, we infer rapid adaptation to P. vivax malaria in humans from the islands of Cabo Verde. Then, we consider the broader spectrum of malaria parasites across primates to begin to ask why some impact human evolution and disease burden more than others. To do this, we will first need new population-genetic methods to interpret patterns of variation in malaria parasites given their complex lifecycles.
January 27, 2022	Rommie Amaro	In Situ Dynamics Reveal Unseen Vulnerabilities of Viral Glycoproteins Viral glycoproteins are important targets for vaccine and drug design. I will discuss our efforts to provide never-before-seen views of these amazing molecular machines, for both SARS-CoV-2 and influenza. For the latter, we investigated the dynamics of influenza glycoproteins in a crowded protein environment through mesoscale all-atom molecular dynamics simulations of two evolutionary-linked glycosylated influenza A whole-virion models. Our simulations reveal and kinetically characterize three main molecular motions of influenza glycoproteins: NA head tilting, HA ectodomain tilting, and HA head breathing. From an immunogenic point of view, the glycoproteins’ accentuated conformational plasticity observed in our simulations has unveiled their full vulnerabilities, exposing epitopes that otherwise would not be accessible or would be sterically occluded. Our work therefore highlights the importance of exploring and characterizing in situ dynamics of proteins in their native, crowded environments.
February 10, 2022	Wah Chiu	CryoEM Informs Protein and RNA Folding Cryogenic electron microscopy (cryoEM) has been advanced to resolve atomic structures of biochemically purified macromolecules with details equivalent to X-ray crystal structures. A unique aspect of cryoEM is to use image processing methods to sort out images of particles with heterogeneous compositions and conformations. This allows us to visualize structures of macromolecules that exist in an ensemble of biochemical states. I will illustrate this approach to show how tubulin is folded within the chamber of human chaperonin TRiC and how ribozyme is transformed from misfolded states to folded state.
February 17, 2022	Philip Romero	Machine learning to navigate sequence-function landscapes for protein engineering Artificial intelligence and machine learning are revolutionizing protein science and engineering by decoding the complex inner workings of proteins at a scale and resolution beyond human comprehension. Predictive sequence-function models enable protein engineers to search for new and useful proteins with broad applications in medicine, bioenergy, biocatalysis, and biotechnology. In this talk, I will present my group’s work leveraging deep neural networks to understand the relationships between protein sequence, structure, and function. We have developed supervised learning methods that infer the sequence-function mapping from large-scale experimental data. These models can be applied to extrapolate beyond the training data to design highly active protein variants. I will also discuss our recent work developing fully autonomous “self-driving labs” that combine AI-based decision making and robotic automation to engineer proteins without human intervention. Data driven protein engineering will become increasingly powerful with continued advances in artificial intelligence, deep learning, and high-throughput and automated experimentation.