Enabling biological discovery through innovations in imaging and computation
The Oldenbourg Laboratory at MBL is developing a new type of light microscope, the LC-PolScope, which uses light polarization to gain insights into submicroscopic structural parameters and non-invasively create contrast where native structures are otherwise invisible. Development of analysis techniques for this instrument requires a concerted effort between experts in physical optics, computational image analysis, and biology. Currently, the complex computations required to process the image data are run on the local computer connected to the LC-PolScope, creating an acquisition bottleneck as these machines lack the computational power to quickly perform these operations. Access to DaLI resources via a high-bandwidth network will allow users to process the images during acquisition, and a shared project space on DaLI will serve as a collaborative platform that will facilitate the development of novel image analysis techniques.
Cell polarization and tissue morphogenesis from molecular interactions
The Munro Laboratory combines quantitative microscopy with detailed agent-based computer simulations to explore how cell polarization and tissue morphogenesis emerge from molecular interactions. Because individual simulations are expensive to run, the ability to store detailed simulation outputs, which can later be “mined” for insights, is critical. However, limitations on data storage and retrieval prevent the long-term storage of these outputs. DaLI will provide the necessary storage to enable these researchers to expand their research towards larger-scale quantitative experimental and computational analysis of cell and tissue dynamics.
The Role of Gesture in Word Learning
Professors Goldin-Meadow and Brentari are interested in how sign language, gestures from hearing and deaf communities, and informal languages such as homesign systems play a role in the acquisition of language. Over several decades, they have collected an extensive video data of 100 children aged between 14 months through 10 years. They captured naturalistic interactions at home and detailed motion capture videos of actions. Organizing and annotating these complex datasets is a major undertaking. In a collaboration with the Center for Gesture, Sign, and Language, the RCC has developed cyberinfrastructure to support the management and distribution of this data in a centralized Sign and Gesture Archive (SAGA). Through DaLI, a server-side data flow can be created to allow users to annotate videos on the server, perform data analysis, and collaborate and share the expected 42TB of data far more efficiently. Data in the Brentari lab is generated from a PhaseSpace 8-camera IMPULSE system.
The first analysis of the role of serotonin in regulating behavioral quiescence and locomotion of the nematode C. elegans in a state-dependent manner
Dr. Biron and collaborators record videos to track the behavior of the nematode C. e/egans over 12-24 hour periods. This increased imaging time allows them to dissect complex behavioral patterns such as lethargus (a sleep-like behavior). Their experiments produce 100-200GB of raw data per day of imaging session. A major limitation on the experimental throughput stems from the problems of storage, access, and backup of about 20TB of raw data per year. The researchers plan to scale up their data acquisition to 100TB per year, but lack the infrastructure to manage such large volumes of data. DaLI will enable this scaling by integrating their acquisition instrument directly with storage space on DaLI.
Topological Vortex Dynamics
Irvine Lab performs three-dimensional visualization at high speeds and high resolution to study the dynamics of linked and knotted vortices in water. In the next three years they predict to generate 150TB of video data. Processing of these videos is time-consuming and often creates a bottleneck that slows acquisition. DaLI would provide the resources to integrate these cameras with HPC resources to carry out computationally and data intensive tasks significantly faster and meet the storage needs for the newly generated data.
Development of Compact, Fast Waterproof Hyper-Spectral Imager & Multi-Channel SpectroPolarimeter for Marine Studies of Coloration and Patterning
The Hanlon Laboratory at MBL studies Rapid Adaptive Coloration in cephalopods and fishes. Dr. Hanlon’s group has developed an innovative light-field sensing instrument that enables dynamic characterization of light in aquatic and other moist environments. The Hanlon laboratory has no formal large-scale storage capacity, and they currently store 19TB of data across multiple desktop drives, with a projection to add an additional 2-3TBper year. DaLI will provide this project with comprehensive data storage resources as well as tools to facilitate the transfer of data from this novel instrument in the field to secure storage on DaLI.
Integrative analysis of ingestive biomechanics and dental microwear in evolutionary and ecological context and Integrative Investigation of the Evolution and Biomechanics of Mandibular Form in Hominids
The Ross Laboratory studies the organization, function, and evolution of vertebrate feeding systems using in vivo measurement of 3D jaw kinematics, muscle activity, bone strain, and cortical neuron activity, combined with computational methods for modeling bone deformation, muscle architecture dynamics, and motor control. One project, that examines the role of primate sensorimotor cortex in control of ingestion, chewing, and swallowing, involves collection of up to 16 channels of physiological data, synchronously with video and neural data. Recent NSF MRI funds have allowed Dr. Ross to acquire a Biplanar Digital Videofluoroscopy Instrument for X-ray Reconstruction of Moving Morphology (XROMM), allowing digital video to be included in this dataset. To facilitate the acquisition and analysis of these datasets, the RCC has worked with the Ross lab to develop the XROMM Data Management Software(XDM). The XROMM equipment was acquired through an NSF MRI grant. The XDM software guides users through an interface that selects experimental files for upload, automatically extracts useful instrument metadata, then stores those files in a key/value store that organizes the data and can be used to view and query data, populate a MySQL database, and automatically upload data to a national repository (the XROMM database at Brown University).
Inaugural Neuroimaging Workshop on the Dynamic Social Brain
The Center for Cognitive and Social Neuroscience is a resource for more than 50 faculty members across more than a dozen departments at UChicago. Research in cognitive and social neuroscience relies heavily on imaging technologies, which yield multi-GB datasets that take countless hours of processing for each participant and scan. Scientists at the RCC and the Center have developed the Chicago Electrical Neuroimaging Analytics (CENA) toolbox, a working HPC prototype for electrical and neuroimaging processing that greatly accelerate this process. For the purposes of data provenance, each processing step and dataset iteration are saved, resulting in up to 1 TB of data per study. Consequently, the Center’s dedicates considerable effort to the storage and analysis of these structural, electrical, and functional brain imaging data. The Center seeks to accommodate resources for the storage and analysis of 75-100 studies in the coming 3 years. DaLI will provide much needed storage space and data management resources for these datasets. The use of the XDM software tools developed at RCC for the Ross lab will facilitate the management of interdisciplinary datasets.
How large neuronal ensembles in the cortex act together to control, coordinate, and learn complex movements of the arm and hand
Hatsopoulos and collaborators work to understand how large neuronal ensembles in the cortex act together to control, coordinate, and learn complex movements of the arm and hand. Using multi-electrode arrays, each of which is composed of 100 microelectrodes, correlated with XROMM 3D Xray fluoroscopy, a 14 camera Vicon motion capture system for collection behavioral data, and a miniature fluorescence microscopy for fluorescence imaging of neural activity. The researchers generate 10 TB of data per year, and have currently collected >50TB. The XDM software available on DaLI will facilitate the transfer and organization of these datasets while providing essential storage capacity and compute resources for these experiments.
Visual code for natural motion and prediction
The Palmer Laboratory uses new computational techniques to study the visual code for natural motion and prediction. Dr. Palmer has built a new natural motion database by making high-speed, high-pixel-depth recordings of natural scenes. By tracking the neuronal response to these videos the Palmer lab seeks to understand how moving objects are tracked and how this information is used to make predictions. Their current bottlenecks involve computation and analysis of data from more than 100 simultaneously recorded neurons, and they would like to expand this to about 1000 concurrent neurons to run real-time clustering during electrophysiology experiments. DaLI will provide tools for tighter integration of these electrophysiological instruments with HPC resources and resources for making the natural motion database available to the broader research community.
Computational investigation of molecular phenomena underlying experimental electrochemical characterization of polymer electrolytes
Juan de Pablo, Paul Nealey, and Shrayesh Patel’s joint project proposes the use of an ensemble of molecular simulation techniques to investigate phenomena governing ion transport behavior in various polymer electrolyte systems. The simulations will be used to inform upon the molecular basis of existing experimental characterizations as well as to guide future synthesis and characterization of polymer electrolyte materials.
