Funded Awards

A thorough understanding of the complexities of the brain’s different cell types requires the sharing and integration of myriad genomic information generated from various data sources. Owen White proposes creating a Neuroscience Multi-Omic (NeMO) Archive, a cloud-based data repository for -omic data. White and his team of researchers will establish an archive for multi-omic data and metadata of the BRAIN Initiative. The group will document and archive data processing workflows to ensure standardization, as well as create resources for user engagement and data visualization. The NeMO Archive will provide an accessible community resource for raw -omics data and for other BRAIN Initiative project data, making them available for computation by the general research community.

Advances in microscopy and imaging have created new possibilities in many fields of research, but these advances have also generated large amounts of data that can overwhelm traditional data management systems. Along with collaborators at Carnegie Mellon University and the University of Pittsburgh, Alexander Ropelewski plans to establish a BRAIN Imaging Archive that takes advantages of infrastructure and personnel resources at the Pittsburgh Supercomputing Center. The Archive will include a pipeline for data submission, user access and support, and BRAIN Initiative community engagement through an online presence, workshops, and hackathons. This unique resource will provide an accessible and cost-effective way for the research community to analyze, share, and interact with large image datasets of the BRAIN Initiative.

The Human Connectome Project provides one of the largest publicly available datasets of diffusion MRI from a sample of healthy individuals. Dr. Rokem and team will create an end-to-end pipeline for analysis of human white matter connections by using “tractometry” methods to analyze the diffusion MRI dataset from the Human Connectome Project. In tractometry, tissue properties are estimated in the long-range connections between remote brain regions. This project aims to generate a normative distribution of tissue properties in the major white matter connections, develop novel statistical methods to connect the properties of white matter connections to cognitive abilities, and create visualization tools to further explore and communicate the data. These tools may create an easily accessible platform that could be applied to other important neuroscience datasets.

The community’s need for an integrated open-source analysis platform is rapidly growing due to the increasing capacity of extracellular electrodes and the limited number of new and validated spike- sorting methods. JRCLUST, a free, open-source, standalone spike sorting software, offers a scalable, automated and well-validated spike sorting workflow for analysis of data generated by large multielectrode arrays. The software can tolerate experimental recording conditions from behaving animals, and it can handle a wide range of datasets using a set of pre-optimized parameters making it practical for wide use in the community. JRCLUST has been adopted in 20+ labs worldwide since its inception less than a year ago. Drs. Kimmel and Nathan seek to expand and maintain JRCLUST, thus empowering researchers to elucidate how functionally defined subpopulations of neurons mediate specific information-processing functions at key moments during behavior.

Non-invasive brain stimulation methods, like transcranial magnetic stimulation (TMS), transcranial direct current stimulation (tDCS), and transcranial alternating current stimulation (tACS), modulate brain activity in a safe and painless manner. Several BRAIN Initiative projects are combining these non-invasive brain stimulation methods with neuroimaging methods. Dr. Opitz and team will develop a computational tool that integrates empirically validated finite element method (FEM) models with imaging data, by extending and scaling the SimNIBS (Simulation of Non-Invasive Brain Stimulation) software platform. The SimNIBS is the leading open source software platform for generating FEM-based models of the electric field distribution produced by TMS, tDCS, and tACS. The resultant analysis package will allow rapid visualization and analysis of non-invasive brain stimulation (NIBS) BRAIN Initiative data, which could be more broadly used by the general neuroscience community.

The proliferation and heterogeneity of magnetic resonance imaging (MRI) experiments, data analysis pipelines, and statistical modeling procedures presents a challenge for effective data sharing and collaboration. Russell Poldrack and colleagues propose expansion of the Brain Imaging Data Structure (BIDS), which standardizes the description and collection of imaging data/metadata for MRI, with development plans for other neuroimaging types as well. Under BIDS, the group will develop standards for pre-processing data pipelines, computational modeling results, and statistical modeling, using quick validation of any implemented standard so that researchers can assess whether their data fit within BIDS guidelines. These standardization goals will facilitate sharing of data, modeling, and results, ensuring their usability and engaging the greater research community in developing highly useable data standards.

Technological advancements in high-resolution imaging of brain volumes permits the accumulation of huge quantities of data that requires solution for storage and archiving. Dr. Brock’s project develops an open, accessible, and cloud-based data archive for electron microscopy and X-ray microtomography data by leveraging the proven architecture of the existing BossDB database. Allowing for petabyte scale data storage, curation, sharing, visualization and analysis, the archive is scalable and allows for a fast in- memory spatial data store, seamless migration of data between low cost and durable object storage (i.e. S3), and rapid access to the enormous datasets. The system enables computing data quality metrics on large datasets and metadata stores through a standardized interface. The archive is developed through an agile process that actively folds in community stakeholders for regular reviews and continuous opportunities for design input.

Dr. Makeig et al. have identified the need for the creation of data archives and standards for specifying, identifying, and annotating the data deposited. Specifically, they aim to create a gateway from the OpenNeuro.org archive for human neuroelectromagnetic data such as EEG and MEG data. This gateway will also provide tools to users for quality evaluation and data visualization. This resource will further allow machine learning methods to be applied to human brain activity data.

AFNI (Analysis of Functional NeuroImages) is an open-source software package for neuroimaging analysis and visualization of both functional and structural MRI as well as other modalities. Drs. Cox and Nielson propose to extend this widely used software package by offering containerization, cloud accessibility and web-accessible visualization. The software extension could support evolving BRAIN Initiative standards for human neuroimaging data organization and experiment specification. The project makes it possible for public integration testing of the software package, thus enabling end-user feedback and wider adoption and dissemination within the neuroimaging community.

Novel neuromodulation, recording, and imaging techniques applied to human and non- human primate brains generate datasets that require tools for organizing, processing and analyzing data that are widely available and easy to use. Drs. Milham and Craddock plan to extend C-PAC (Configurable Pipeline for the Analysis of Connectomes), building a configurable data analysis pipeline that incorporates various statistical analysis, machine learning, and network analytic techniques. In addition to adapting methods used in human imaging for non-human primate data, the project will implement a toolbox for alignment of electrophysiological data with brain imaging data. The resulting software enables high- throughput, semiautomated and end-to-end processing and analysis of structural and functional MRI data that are accessed locally or via the cloud.

Recent tissue-clearing techniques and advances in microscopy have made it possible to produce 3D images of intact brains. to help ensure consistency in data collection and analysis, Dr. Hamilton and her team will develop a set of standards for3D imaging of whole brains for the neuroscience research community.. Dr. Hamilton’s group will convene a Working Group of experts who will work through a consensus process to establish standards that will be distributed to the research community. These standards should help improve the efficiency of imaging research and allow comparisons across studies.

Dr. Ghosh and colleagues propose to develop a new computer infrastructure for scientists to share, collaborate, and process neurophysiological data. The data will be open to both scientists and also be used to engage high school and college students. The team will also develop tools to facilitate data submission and access and to encourage adoption of the Neurodata Without Borders data standard.

This project develops DABI (Data Archive for the Brain Initiative) to aid the dissemination of human neurophysiological data generated through the BRAIN Initiative. Incorporating infrastructure from a pre- existing hub for delivering effective informatics and analytics solutions for major projects in the study of neurological diseases, Drs. Toga, Duncan, and Pouratian will aggregate data related to human electrophysiology, making the data broadly available and accessible to the research community. The group plans to incorporate analysis tools with user interfaces, implement tools for data management and use, and link metadata across different data modalities. The overarching goal of this project is to secure, link, and disseminate BRAIN Initiative data with all pertinent recording and imaging parameters coming from participating sites

Many labs develop unique software to manage and interpret their findings, but those programs are often specific for certain types of datasets, making them difficult to share among researchers. Dr. Van Hooser’s team plans to create an interface standard that establishes a common set of processes for accessing neurophysiological and imaging data. The standard will be tested, and revised accordingly, based on feedback from graduate students and postdoctoral researchers during data access events, or “hack-a-thons.” The interface standard will help increase the speed of research and make data widely available, allowing individuals outside of the neuroscience and research communities to make discoveries.

Different laboratories use different behavioral systems, hardware, and software, which makes it difficult to standardize, communicate, and replicate experiments. Dr. Kepecs et al. have led the creation of a laboratory consortium to address this problem by developing requirements for behavioral data formats and an open source software suite for editing, executing, and visualizing behavioral tasks. They will also engage the scientific community to promote the adoption of these standards and tools.

Molecular and cellular neuroscientists often lack the training in computer programming to fully explore “-omics” data common in the BRAIN Initiative. Drs. Hertzano and White will implement analytic software for visualization and interactive genome browsing of gene expression and RNA-seq data, including simple and complex cross-dataset analysis. These tools will be made available in the BRAIN Initiative funded Neuroscience Multi-Omic Data Archive (NeMO) which hosts multi-omic data. The software will provide an easy-to-use web-based work environment for visualization and analysis of multi-modality and multi- omic data, interrogation of relationships between epigenomic signatures and gene expression, and integration of analytical techniques for multivariate analysis, gene co- expression and other analyses.

A thorough understanding of brain function requires the integration of neuroscience data across species and scales. While current software can verify data quality within one or a handful of data sources, reproducibility across multiple data sources is limited. Xenophon Papademetris and colleagues are developing software tools with cross-scale, cross-species reproducibility analysis in mind. By leveraging data created by two other BRAIN Initiative projects at Yale University, Papademetris will extend current software algorithms to incorporate data from multiple sources, design the software to be cross-platform compatible, validate the software through rigorous testing, and finally, distribute it to the community. The potential for a set of software tools to reliably and reproducibly analyze multiple heterogeneous neuroscience data types will help to break down data barriers for the greater neuroscience community.

Neurophysiology research, which focuses on recording brain cell activity, produces enormous amounts of complex data that are difficult to manage. Drs. Rubel and Ng will build upon the Neurodata Without Borders: Neurophysiology project to create a system that will allow for standardizing, sharing, and reusing neurophysiological data. The team will design an open source software system; develop methods to establish a consistent vocabulary for defining cell types, measurements, and behavioral tasks; and create tools to help the community adopt these new resources and standards. The proposed system will help accelerate neurophysiological discoveries as well as reproducibility studies.

To leverage the public investment in the BRAIN Initiative, the sharing of data produced by its myriad projects is paramount. Dr. Podrack’s project extends the recently released OpenNeuro, which was developed based on the well-established and successful OpenfMRI, for an archive of neuroimaging data. The extended archive encompasses a broader range of neuroimaging data including EEG, MEG, diffusion MRI and others. The archive also implements easy-to-use data submission, semi-automated curation and advanced data processing workflows, which run directly on the cloud platform. The archive allows to share the results alongside the data, federate with other relevant repositories, and accessible to all researchers.

Neuroimaging techniques are advancing at a rapid rate, resulting in high resolution images of brain tissue and large datasets that can be difficult to manage. William Gray Roncal’s team propose an integration framework called SABER: Scalable Analytics for Brain Exploration Research. With a focus on techniques (e.g., electron microscopy) that produce high resolution brain images, SABER will create a unified framework by which these data types are accessible to a broader audience, can be processed in a reproducible, portable way, and can be scaled from small data volumes to large datasets. SABER has the potential to enable discoveries from high-resolution imaging techniques, making the parsing of entire brains a new reality.

The characterization of single neurons is a major theme of research in the BRAIN Initiative Cell Census Network. Dr. Mukamel and team develop Single Neuron Analyzer (SNA), a cloud-based computer tool for interactive data exploration, analysis, and validation for single cell molecular data generated by the BRAIN Initiative. The software specifically implements machine learning and cross-validation techniques to improve the reproducibility of neuronal cell type identification.

Inconsistent terminologies of neuroimaging metadata prevent the precise description of the design and intent of an experiment, experimental subject characteristics, and the data acquired. Dr. Keator and colleagues aim to address this problem by developing human neuroimaging domain-specific controlled vocabularies. This will greatly improve our ability to search across datasets; to reuse, compare and integrate data across studies and sites; and to replicate neuroscience findings. They also aim to promote the adoption of the controlled vocabularies through community engagement.