Michael Cianfrocco

Michael Cianfrocco

By |

Dr. Michael Cianfrocco uses cryo-electron microscopy (cryo-EM) to determine protein structures to understand how nanometer-sized molecular machines work. While a powerful technique, cryo-EM data collection and subsequent image analysis remain bespoke, clunky, and heuristic. Dr. Cianfrocco is coupling his 16+ years of experience with artificial intelligence to automate data collection and processing by capturing human expertise into AI-based algorithms. Recently, his laboratory implemented reinforcement learning to guide cryo-electron microscopes for data collection [1, 2]. This work combined real-world datasets and Dr. Cianfrocco’s expertise with AI-driven optimization algorithms to find the ‘best’ areas of cryo-EM samples for data collection.

cryoRL Distributed Data Collection process diagram

Human users must curate and select areas for subsequent analysis after data collection. Subjective decisions guide how to process the single particles and determine what constitutes ‘good’ data. To automate subsequent preprocessing, Dr. Cianfrocco’s lab built the first AI-backed data preprocessing in cryo-EM by training CNNs to recognize ‘good’ and ‘bad’ cryo-EM data [3]. This work enabled fully-automated cryo-EM data preprocessing, the first step in the processing pipeline of cryo-EM data. In the future, Dr. Cianfrocco wants to continue improving cryo-EM workflows to make them robust and automated, eventually surpassing human experts in the ability of algorithms to collect and analyze cryo-EM data. 1. Fan Q*, Li Y*, et al. “CryoRL: Reinforcement Learning Enables Efficient Cryo-EM Data Collection.” arXiv preprint arXiv:2204.07543 (2022). 2. Li Y*, Fan Q*, Optimized path planning surpasses human efficiency in cryo-EM imaging. bioRxiv 2022.06.17.496614 (2022). 3. Li Y, High-Throughput Cryo-EM Enabled by User-Free Preprocessing Routines. Structure. 2020 Jul 7;28(7):858-869.e3.

Peter Song

Peter Song

By |

My research interests lie in two major fields: In the field of statistical methodology, my interests include data integration, distributed inference, federated learning and meta learning, high-dimensional statistics, mixed integer optimization, statistical machine learning, and spatiotemporal modeling. In the field of empirical study, my interests include bioinformatics, biological aging, epigenetics, environmental health sciences, nephrology, nutritional sciences, obesity, and statistical genetics.

Anthony Bloch

Anthony Bloch

By |

My research interests include : Hamiltonian and Lagrangian mechanics, gradient flows on manifolds, integrable systems stability, the motion of mechanical systems with constraints, the relationship between continuous and discrete flows, nonlinear and optimal control and the control of quantum systems. I also interested in data-guided control and in particular the dynamics and control
of networks and systems arising from large sets, particularly in biological applications.

Ryan Stidham

Ryan Stidham

By |

Dr. Stidham is an academic gastroenterologist specializing in medical image analysis in Crohn’s disease, ulcerative colitis, inflammatory bowel diseases (IBD), and gastroenterology conditions at large. His research is focused on developing new measures of disease activity to power automated care models and clinical decision support systems in IBD with a focus on medical image analysis and new technology development. His work has focused on automation of existing IBD disease measures that relying on colonoscopy, CT, MRI, and ultrasound using neural networks and novel image analysis approaches. Dr. Stidham is also developing new measures of disease activity, inflammation, and fibrosis that leverage advances in image segmentation, transfer learning, signals analysis, and fuzzy network approaches as well as collaborating for development of new image acquisition modalities. Finally, his team has active projects in collaboration with the Department of Learning Health Sciences for merging data from clinical office notes with imaging data using computational linguistics approaches. His work has been supported by the NIH, DOD, NSF, and several large investigator-initiated industry collaborations.

Dan Rabosky

Dan Rabosky

By |

The Rabosky lab seeks to understand how and why life on Earth became so diverse. We focus primarily on large-scale patterns of species diversification (speciation and extinction) and on the tempo and mode of phenotypic evolution, to better understand what regulates the “amount” of biodiversity through Deep Time. To this end, we develop theoretical frameworks and computational tools for studying evolutionary dynamics using DNA-sequence-based evolutionary trees (phylogenies), the fossil record, as well as phenotypic data from present-day species (morphology, ecology). We develop and apply a range of methods involving supervised and unsupervised learning, including Markov chain Monte Carlo, hierarchical mixture models, hidden Markov models, latent feature models, and more. We are increasingly interested in complex morphological and ecological traits, which – due to a rapidly expanding data universe – represent a tremendous opportunity for the field to answer long-standing questions about how organisms evolve. At these same time, we are embracing the analytical challenges of these data, because fully realizing their potential requires the development of new analytical paradigms that go beyond the limitations of traditional parametric models for low-dimensional data.

Automatic feature identification from a large-scale evolutionary tree (phylogeny) using a compound model of the generating process (speciation, extinction) developed in the Rabosky lab. Colors correspond to distinct evolutionary rate regimes as estimated using Markov chain Monte Carlo. This method revealed widespread heterogeneity in the rate of species formation during 350 million years of ray-finned fish evolution. Warm colors = fast rates; cool colors = slow rates.

Automatic feature identification from a large-scale evolutionary tree (phylogeny) using a compound model of the generating process (speciation, extinction) developed in the Rabosky lab. Colors correspond to distinct evolutionary rate regimes as estimated using Markov chain Monte Carlo. This method revealed widespread heterogeneity in the rate of species formation during 350 million years of ray-finned fish evolution. Warm colors = fast rates; cool colors = slow rates.

Photograph of Alison Davis Rabosky

Alison Davis Rabosky

By |

Our research group studies how and why an organism’s traits (“phenotypes”) evolve in natural populations. Explaining the mechanisms that generate and regulate patterns of phenotypic diversity is a major goal of evolutionary biology: why do we see rapid shifts to strikingly new and distinct character states, and how stable are these evolutionary transitions across space and time? To answer these questions, we generate and analyze high-throughput “big data” on both genomes and phenotypes across the 18,000 species of reptiles and amphibians across the globe. Then, we use the statistical tools of phylogenetic comparative analysis, geometric morphometrics of 3D anatomy generated from CT scans, and genome annotation and comparative transcriptomics to understand the integrated trait correlations that create complex phenotypes. Currently, we are using machine learning and neural networks to study the color patterns of animals vouchered into biodiversity collections and test hypotheses about the ecological causes and evolutionary consequences of phenotypic innovation. We are especially passionate about the effective and accurate visualization of large-scale multidimensional datasets, and we prioritize training in both best practices and new innovations in quantitative data display.

Photograph of Nate Sanders

Nate Sanders

By |

My research interests are broad, but generally center on the causes and consequences of biodiversity loss at local, regional, and global scales with an explicit focus on global change drivers. Our work has been published in Science, Nature, Science Advances, Global Change Biology, PNAS, AREES, TREE, and Ecology Letters among other journals. We are especially interested in using AI and machine learning to explore broad-scale patterns of biodiversity and phenotypic variation, mostly in ants.

Picture of Besa Xhabija

Besa Xhabija

By |

Dr. Xhabija joined the Department of Natural Sciences in September 2022 as an Assistant Professor of Biochemistry. Her laboratory aims to understand the effects of toxins on early embryonic development utilizing embryonic stem cells because they provide a new tool and opportunity to investigate the impact of environmental exposures and their interactions with genetic factors on human development and health. To fully realize these potentials, she believes that it is important to understand the molecular basis of the defining characteristic of the stem cells. More specifically, she is interested in investigating how stem cells play a role in shaping the expression program during development and how mechanisms of self-renewal and differentiation during mammalian development regulate cellular fate decisions.

Picture of David Brang

David Brang

By |

My lab studies how information from one sensory system influences processing in other sensory systems, as well as how this information is integrated in the brain. Specifically, we investigate the mechanisms underlying basic auditory, visual, and tactile interactions, synesthesia, multisensory body image perception, and visual facilitation of speech perception. Our current research examines multisensory processes using a variety of techniques including psychophysical testing and illusions, fMRI and DTI, electrophysiological measures of neural activity (both EEG and iEEG), and lesion mapping in patients with brain tumors. Our intracranial electroencephalography (iEEG/ECoG/sEEG) recordings are a unique resource that allow us to record neural activity directly from the human brain from clinically implanted electrodes in patients. These recordings are collected while patients perform the same auditory, visual, and tactile tasks that we use in our other behavioral and neuroimaging studies, but iEEG measures have millisecond temporal resolution as well as millimeter spatial precision, providing unparalleled information about the flow of neural activity in the brain. We use signal processing techniques and machine learning methods to identify how information is encoded in the brain and how it is disrupted in clinical contexts (e.g., in patients with a brain tumor).

Krishna Garikipati

Krishna Garikipati

By |

My research is in computational science and scientific artificial intelligence, including machine learning and data-driven modelling. I have applied these approaches to physics discovery by model inference, scale bridging, partial differential equation solvers, representation of complexity and constructing reduced-order models of high-dimensional systems. My research is motivated by and applied to phenomena in bioengineering, biophysics, mathematical biology and materials physics. Of specific interest to me are patterning and morphogenesis in developmental biology, cellular biophysics, soft matter and mechano-chemical phase transformations in materials. More fundamentally, the foundations of my research lie in applied mathematics, numerical methods and scientific computing.

A schematic illustrating the range of ML methods comprising the mechanoChemML code framework for data-driven computational material physics.