Partha Pratim Mitra - US grants

Temporal studies were performed both to understand spatiotemporal characteristics of signal in fMRI images and to examine temporal evolution of neuronal processing. In a time-resolved study of the visual system, the nature of the spatiotemporal characteristics of functional magnetic resonance imaging was examined using sophisticated mathematical tools. This study revealed a rich spatiotemporal structure in the stimulus related changes even for a very simple visual stimulation protocol, suggesting that distinct structures respond differently to the stimulus. In addition to revealing the temporal characteristics, this work also presented a number of novel mathematical approaches for processing the data, including a novel approach for removing gross motion artifact.

DESCRIPTION (Adapted from Applicant's abstract): The aim of this research project is to determine the temporal correlation structure of neuronal activity within and between local and distant cortical areas in a fronto-parietal network during working memory tasks. The research will be performed via simultaneous multiple tetrode recordings from the relevant cortical areas in macaque monkeys performing saccade and reach memory tasks. Multivariate spectral analysis tools will be used to analyze and interpret the neuronal activity. These tools were applied to the data collected during preliminary recording from parietal area LIP during a memory saccade task and revealed a spatially tuned increase in power spectrum in both single unit and local field potentials between 40-90 Hz during the memory period. This is the first demonstration of temporal coding during a memory task. The specific aims will test hypothesis relating to correlations in temporal structure in both single units and local field potentials activity (as compared to the spike rates). The first aim will complete the preliminary study testing the hypothesis that SU and LFP spectra in area LIP show memory fields in 40-90 Hz band by recording from multiple tetrodes during a memory saccade task and characterizing the correlation between SU and LFP activity. The second aim will test the hypothesis that the activity in the FEF in prefrontal cortex and also activated by a memory saccade task, show similar dynamic memory fields of SU and LIP spectra. This will be tested by recording from multiple tetrodes in FEF during the same task. The third aim will test the hypothesis that the temporal structure during working memory is correlated across areas LIP and FEF using multiple tetrode recordings from these areas. The fourth aim will test the hypothesis that the temporal structure and correlation observed in areas LIP and FEF are reflected by analogous activity in reach circuitry during a memory-reach task. This hypothesis will be tested using data recording from parietal reach region (PPR) and dorsal premotor area (PMd) following methodology of the first three aims. Elucidating the organization of temporal structure and correlations in neuronal activity during a working memory task is necessary to determine the underlying functional architecture. The instrumental and analytical techniques developed in this proposal, particularly in the context of LFP, have fundamental significance for development of neural prosthetics. The experiments will promote understanding of short-term memory, which is often affected with aging and neurological disease.

DESCRIPTION (provided by applicant): The goal of the Neuroinformatics course at the Marine Biological Laboratory is to train 25 researchers a year, ranging from advanced students to senior investigators, in statistical and computational techniques for the analysis of a broad spectrum of neuroscientific data. This is a natural outgrowth of the MBL Workshop on Analysis of Neural Data, which helped crystallize the community and intellectual agenda reflected in this course.The course spans two weeks, and by emphasizing data analysis and informatics is complementary to the Methods in Computational Neuroscience course at the MBL, which focuses on modeling and theory. The first week consists of pedagogical lectures in statistics, ranging from an elementary introduction to advanced topics. Evenings will have tutorials to help attendees gain experience and concrete understanding. There will also be lectures on data acquisition techniques, stimulus design and the underlying neurobiology. The second week will include focused workshops on various topics, such as neural prostheses, temporal coding, functional MRI. Participants will present their research throughout the course, and analyze their own data in the open laboratory time. All faculty members are expected to interact closely with the attendees and with each other, to analyze specific data sets and to advance the field in general.The course is unique in the spectrum of subjects covered in a unified manner, ranging from multiple spike trains to functional imaging data, and ranging from neuroscience to biomedical engineering. To assist the development of standards in this nascent field, the course presents a unified pedagogical approach grounded in the relevant disciplinary areas of statistics, mathematics and physics, while maintaining intimate ties to neurobiological and biomedical problems. The interdisciplinary scope of the course cannot currently be replicated in a university setting, although the syllabus and experience gathered in this course will prove invaluable for future university courses. Apart from the academic implications, the course has clear consequences for advanced medical technology, and helps involve physical and computer scientists in neurobiology.

DESCRIPTION (provided by applicant): Neuroscientists acquire single-channel, multichannel, and spatiotemporal data in ever-increasing quantities and need high-quality software tools for preprocessing, exploration, and analysis. Such tools are critical to addressing pressing neuroscientific questions, such as the nature of the neural code and the role of correlated activity of many neurons in perception and behavior. While several groups have developed related tools for internal use, there is currently no software system suitable for multiple data types and formats, usable for large volumes of data, with high-quality numerics, and professionally documented and maintained. Such a software system would enhance neuroscience research by (a) providing access to advanced analytical tools to groups that lack the resources to create them internally, (b) reducing the duplication of effort across groups, (c) allowing for a fuller utilization of physiological data sets, and (d) improving the communication of results between groups. Over the past five years, precisely such a set of tools have been developed by the PI and colleagues. These are in use in over 24 laboratories worldwide for research, in the Neuroinformatics course at the Marine Biological Laboratories for pedagogy, and have led to over 25 publications. In response to many requests, demonstrated by the approximately 24 appended letters, the aim of this grant is the continued development, maintenance and distribution of Chronux, an open source software package for advanced analysis of neurobiological time series data. Specific aims are: 1. Further development and maintenance of a numerical analysis library. Initial emphasis will be on electrophysiological data, both spikes and continuous signals (EEG/MEG/LFP), but later releases will provide routines specialized for image time series data. 2. Incorporation of an I/O library for multiple neurobiological data and metadata formats, consisting of a set of filters to a number of existing formats, along with a general specification for such filters. 3. Development of a graphical user interface, to facilitate use by the typical neuroscientist. 4. Documentation, distribution, maintenance and quality assurance for Chronux. We will work closely with approximately 24 laboratories and the Neuroinformatics course at the MBL via a feedback process to ensure the quality and usability of Chronux.

ABSTRACT
0709983
Mitra, Partha
Cold Spring Harbor Lab
Engineering Principles in Bioogical Systems

Three workshops in a series entitled Engineering Principles in Biological Systems are proposed to take place at Cold Spring Harbor Laboratory's Banbury Center, Cold Spring Harbor, NY on May 6-8, 2007, May 4-6, 2008, and May 3-5, 2009. The purpose of these workshops is to bring together scientists with strong theoretical or mathematical backgrounds, and an active interest in applying engineering principles to the study of biological systems, for mutual education and future collaboration. Participants will be drawn from engineering and computer science, biology, and the physical sciences, and their topics of research will span cellular, systems and population biology. Intellectual Merit: Through the process of evolution, living systems retain accidentally found solutions to problems they must solve in order to survive. In the past, theoretical
biology has largely focused on explanations of the physico-chemical mechanisms behind such solutions, while explanation in the form of function-solution pairs has been studied in a relatively ad hoc manner and has not been approached from a disciplinary perspective. These workshops will promote the development of an emerging approach to theoretical biology with more formal emphasis on design or engineering principles. Here, the premise is that although solutions or designs in biological systems are not engineered but instead arise incrementally through natural selection, they may nevertheless be studied in their existing forms in the framework of engineering theories developed alongside human-engineered systems.
Goals of the workshops:
(1) Development of a theoretical canon: Refine a list of theories, rather than a theory of everything, applicable to studying engineering principles in biological systems. A starting point for these theories (as reflected in the proposed workshop sessions) can be drawn from courses taught in engineering departments. The idea is therefore to start with major existing engineering theories (controls, communication, computation) and to examine whether these apply to biological systems, and if not, what modifications are in order.
(2) Pedagogical goals: The workshops will provide an educational opportunity for biological researchers to learn about engineering theories which may be relevant to their work, and for engineering theorists and computer scientists to learn about biological problems they might help to be understood. Each session will have a tutorial overview of the corresponding engineering theory, followed by biological examples, chosen
specifically to span scales: from cellular and organism levels, as well as from population or evolutionary biology where appropriate. Dissemination and Broader Impact: The proposed workshops will foster this new
approach to understanding biological systems and the collaborative culture across disciplines that its success will require. The conference venue and format are ideal for encouraging open discussion and initiating collaborative efforts, and it is hoped that a continuing series of such workshops will also encourage the development of an enduring and progressive theoretical framework. To complement the direct training opportunity for workshop participants, talks presented at the workshop along with relevant supporting
material will be put on a publicly accessible website for the benefit of the general scientific community. In addition, a summary report reflecting the deliberations at the workshop will be prepared and disseminated through the website and submitted for publication in a review journal.

DESCRIPTION (provided by applicant): The goal of this project is exploratory research into localization of distributed sources of neural activity from joint MEG/EEG measurements, using a recently developed methodology for the inverse problem (Local Basis Expansions or LBEX). Neural dynamics in the 10-100 ms timescale underlie a broad spectrum of cognitive function relevant to the study of mental disorders. EEG and MEG are the only practical noninvasive techniques with this time resolution, but are limited in spatial resolution due to the ill posed inverse problem relating sensor measurements to underlying sources. EEG is more widely used, but also suffers from uncertainties about head conductivity profiles. The MEG inverse problem has proven to be comparatively more tractable. However, MEG is substantially more expensive, and it is desirable to combine the two techniques, so that EEG sources can be better pinpointed using simultaneous MEG measurements. Another reason to combine the two methods is the complementary nature of the inverse problems: EEG silent sources can be MEG active, and vice versa. Thanks to recent instrumental advances, simultaneous measurements are now routinely possible, but the joint localization problem has not yet been fully studied. We have developed a methodology for MEG source localization (LBEX), which permits a systematic analysis of resolution limits, through a mathematical treatment of the uncertainty principle governing the inverse problem. In this proposal, we propose to explore extensions of the LBEX method to joint MEG/EEG source localization. Specific aims include (i) a theoretical analysis of the joint localization problem in an idealized spherical model, (ii) evaluation of the performance of the technique in numerical simulations with a realistic head model, and (iii) application of the joint localization technique to simultaneous MEG/EEG recordings obtained through an experimental collaboration. Successful completion of the research program will establish the utility of our algorithmic framework, and will also evaluate the gain obtained from simultaneous MEG/EEG localization. Once the methodology is validated by the exploratory research, we intend in the future to encode the results into user friendly software which will enable widespread usage of joint EEG/MEG localization.

DESCRIPTION (provided by applicant): Brain function is dictated by its circuitry, yet we know little about its wiring architecture: in the most-studied mammal (rat), only an estimated 10-30% of the long range circuit connections have been probed. There is growing consensus that it is time to close this gap by generating brainwide connectivity maps for model vertebrates. The mouse is the starting species of choice: mouse models form the backbone of research into the etiology of neuropsychiatric disorders, have the most-studied genome of any mammal, and are key to the early stages of drug development. Over the last two years, we have organized several meetings involving the neuroanatomy community to gain in-depth understanding of the technical and scientific challenges of such a project. Based on this experience, we propose to produce the first brainwide connectivity map of mouse, through the development of an automated pipeline of experimental and computational techniques-- a connectivity scanner. Our proposal is timely and is enabled by advances in automated wide-field slide scanning microscopy, decreasing data-storage costs, and established tract-tracing methods using injections of classical tracers and neurotropic viruses. The need for brain-wide scope and scalability rule out other approaches. To demonstrate the translational utility of the approach, we will also analyze disease model mice (autism and schizophrenia), to understand alterations in the connectivity map compared to the reference map generated in the project. The project does not fit neatly into an existing funding mechanism at the NIH, but has the potential to fundamentally impact the entire neuroscience community. The transformative potential is twofold: by generating the first mammalian brainwide connectivity map, we provide the neuroscientific community with a landmark reference map which can be used in a wide variety of contexts. Secondly, our emphasis on open source software development, cost optimization and duplicability will result in an affordable, integrated instrument which other academic laboratories will be able to implement. In this way it borrows from, yet differs significantly from, the model offered by the Allen Institute, which has previously demonstrated the potential of industrial automation for neuroscience.

DESCRIPTION (provided by applicant): This application addresses broad Challenge Area (15): Translational Science and specific Challenge Topic, 15- MH-103 Mapping the Neural Connectivity of a Mouse Model. Brain function is dictated by its circuitry, yet we know little about its wiring architecture: in the most-studied mammal (rat), only an estimated 10-30% of the long range circuit connections have been probed. The present Challenge Topic validates the growing consensus that it is time to close this gap by generating brainwide connectivity maps for model vertebrates. Over the last two years, we have organized several meetings involving the neuroanatomy community to gain in-depth understanding of the technical and scientific challenges of such a project. Based on this experience, we have designed and have begun to build and test an automated pipeline of experimental and computational techniques for achieving this goal. Our proposal is enabled by advances in automated wide-field slide scanning microscopy, decreasing data-storage costs, and established tract-tracing methods using injections of classical tracers and engineered viruses. The experimental plan can be summarized as follows. The mouse brain is divided into ~200 regions based on classical neuroanatomical and regional gene-expression data. For each region we inject one mouse with classical tracers and one mouse with viral tracers. From the injection site, the tracers are transported anterogradely to the area's projection targets and retrogradely to areas which project to the injection site. In this way, individual projections are revealed multiple times. In order to acquire this information, we will section the entire brain from each mouse and image the sections using an automated slide-scanning microscope. The resulting 2D slice-images will be combined in software to produce a 3D reconstructed brain image for each injection. Finally the 3D images from all of the individual injections will be combined by spatially registering them to the Allen Reference Atlas, ultimately generating a unified brainwide neural connectivity map. Generating the first unbiased, brainwide connectivity map in the mouse will have broad neuroscientific implications. The study of neural development, neural network modeling, evolutionary neuroanatomy, and associative and integrative brain function will benefit tremendously from finally having this landmark reference map to meaningfully constrain theories and aid in experimental design and interpretation of results. Relationships between gene expression and connectivity can be probed by analyzing the gene-expression maps generated by the Allen Institute in combination with the connectivity maps generated by this project. The baseline neural connectivity map generated in the present study will serve as a foundation for subsequently studying circuit polymorphisms across mutant mouse lines. The ability to objectively quantify alterations in connectivity in mouse models of neuropsychiatric disorders such as autism and schizophrenia will aid our understanding of their etiology and pathophysiology. Finally, our emphasis on open source software development, cost optimization and duplicability will result in an affordable, integrated instrument which other academic laboratories will be able to implement, so that this approach can be rapidly applied to a wide variety of neuroscientific problems. NARRATIVE The study of mouse models of neuropsychiatric disorders provides hope for the development of therapies for these burdensome illnesses, but progress has been slow due to the lack of knowledge about how the mouse brain is wired. This project aims to close this gap by generating the first brain-wide wiring diagram of mouse, automating techniques that are known to work but are labor-intensive. If successful, the project has the potential to fundamentally transform our understanding of the architecture of the normal and disordered brain.

DESCRIPTION (provided by applicant): The increasing availability of genome wide data sets promise to shed light into the etiology and pathophysiology of genetically complex disorders, including substance abuse and dependence. There remain significant challenges, however: although there is evidence for significant heritability, genome wide association studies have typically revealed small effect sizes, possibly due to the polygenic nature of the disorders. The brain-wide gene expression data sets from the Allen Institute offers new data sources that could be used to group genes together based on similarities in their expression profiles in anatomic space, thus enhancing the power of statistical tests in genome-wide studies. Due to the unprecedented spatial resolution in these data sets, with genome-wide and brain-wide coverage, specific hypotheses involving intercellular biochemical networks as well as brain-wide neural networks can also be examined. At the Allen Institute and at Cold Spring Harbor Laboratory, we have been collaboratively analyzing the Allen Brain Atlas (ABA) adult mouse brain data set, and preliminary results demonstrate that the spatial co-expression patterns of genes are indeed a rich source of information. In this application, we intend to focus this analysis on addiction-related gene sets, in consultation with experts on addiction research and integrating relevant online information resources. Specific aims in the first year (R21 phase) include (1) development and refinement of software and web-based tools for analysis of co-expression patterns in gene sets and (2) multivariate analysis of an initial set of addiction related genes. The first year will focus on the adult mouse brain data set that is already at hand. In subsequent years (years 2-4, R33 phase), we will extend the co-expression analysis to mouse developmental and spinal cord data sets (Aim 1), and human brain data sets (Aim 2), that are scheduled to become available during this period. Additionally, we will mine existing databases and the literature to augment our initial gene lists as well as to develop a database of associations between substance abuse phenotypes and corresponding brain areas (Aim 3). This will allow us to more fully analyze the intra and intercellular networks that may be involved in addiction. Finally, we will make the computational tools and analysis results developed as part of our research publicly available in the form of a web portal (aim 4). PUBLIC HEALTH RELEVANCE: The identification of genes and gene networks driving drug abuse and addiction is a major current challenge in addiction genetics. The public presentation of tools for understanding spatially mapped genomic datasets such as the ABA will have major impact on researchers aiming to understand these genetic networks and pathways. The proposed work will encompass both the identification of key addiction gene clusters as well as the generation of useful online methods for addiction researchers.

We propose to provide support for graduate students and postdoctoral researchers to attend a
conference on Engineering Principles for Biological Systems, to be held at Cold Spring Harbor
Laboratory on 10--13 December 2008. The conference is intended to foster cross-disciplinary
exchange of ideas and expertise between engineers, mathematicians and biologists interested
in the analysis of diverse biological systems through the application of engineering principles.
Support from this grant would be used to help offset the cost of the conference for graduate
students and postdoctoral researchers who wish to attend.

Intellectual Merit: Through the process of evolution, living systems retain solutions arisen by
chance to problems they must solve in order to survive. In the past, theoretical biology has
largely focused on explanations of the physico-chemical mechanisms behind such solutions,
while explanation in the form of function-solution pairs has been studied in a relatively ad hoc
manner and has not been approached from a disciplinary perspective. This conference will
promote the development of an emerging approach to theoretical biology with more formal
emphasis on design or engineering principles. Here, the premise is that although solutions or
designs in biological systems are not engineered but instead arise incrementally through natural
selection, they may nevertheless be studied in their existing forms in the framework of
engineering theories developed alongside human-engineered systems.
This conference is part of a series of workshops and conferences organized by Partha Mitra
(CSHL), Richard Murray (Caltech) and others to help foster a community of researchers working
on theoretical frameworks for understanding biological systems across a variety of scales. A
starting point for these theories can be drawn from courses taught in engineering departments.
The idea is therefore to start with major existing engineering theories (controls, communication,
computation) and to examine whether these apply to biological systems, and if not, what
modifications are in order. The conference series and this CSHL conference in particular, will
provide an educational opportunity for biological researchers to learn about engineering theories
which may be relevant to their work, and for engineering theorists and computer scientists to
learn about biological problems they might help to be understood. Each session at the
conference will have a two invited talks, one each by a biologist and a theoretician/engineer,
integrated with a set of contributed talks, chosen from submitted abstracts.

Dissemination and Broader Impact: The proposed conference will foster this new approach to
understanding biological systems and the collaborative culture across disciplines that its
success will require. The conference venue and format are ideal for encouraging open
discussion and initiating collaborative efforts, and it is hoped that a continuing series of such
meetings will also encourage the development of an enduring and progressive theoretical
framework. To complement the direct training opportunity for conference participants, talks
presented at the conference will be put on a website for the benefit of the wider scientific
community. Options will be provided to individual presenters to make their talks publicly
available through the Leading Strand web site.

DESCRIPTION (provided by applicant): Currently in the neuroimaging community (and elsewhere in neuroscience), researchers employ a variety of deferent procedures and atlases to parcellate and label brain regions that are of interest in their work. The result is that many dierent labels are used to indicate the same spatial region, and in some cases, the same label is used to indicate dierent regions. This well-known nomenclature problem, which has the negative consequence of making cross-study comparison especially dicult, has previously been addressed by developing semantic mappings between synonymous and/or hierarchically-related region labels. How- ever, the spatial relationships between regions as delineated by dierent atlases can be quite complex, are poorly understood, and are not adequately captured by such semantic mappings. We have developed a new approach to this atlas concordance problem based on analyzing the spatial relationships between various brain parcellations. By applying our metrics to dierent parcellation schemes now in use, we found that the overall concordance between partitions is rather poor, which suggests the need for a \meta-atlas or systematic procedures for mapping between dierent atlases. Through this grant proposal, we wish to expand upon the tools and methods which we have developed for this purpose, and make them available to the neuroscience community. The rest specic aim will be to make the atlas comparison and meta-analysis tools available online through an interactive, customizable website. The second aim is for algorithmic innovations to enhance the concordance analysis of brain parcellations and nomenclatures. This will include the incorporation of additional atlases and functionality as well as further theoretical developments of overall atlas concordance measures. The third aim is integration with BIRN and caBIG infrastructures. This will include mappings from the brain region labels used in the analyzed atlases to the BIRNLex ontology, additional web services within the BIRN Atlas Interoperability Framework, and adding appropriate functionality to the neuroimage processing pipelines on the BIRN GRID. Successful completion of our project will enhance data integration and meta-analysis of neuroimaging data sets, and broadly impact both basic and clinical research in neurology and neuropsychiatry. PUBLIC HEALTH RELEVANCE: Functional brain imaging techniques have revolutionized basic and clinical neuroscience, but there are some signicant challenges - particularly, a multiplicity of brain atlases and nomenclature schemes that make cross-study comparison and meta analysis dicult. We have developed a framework to deal with this atlas concordance problem, including quantitative measures and software tools. The proposed research will make these tools available to the general neuroimaging community, expand and improve the preliminary analysis and will integrate with the BIRN and caBIG infrastructures.

This INSPIRE award is partially funded by the Physics of Living Systems program in the Division of Physics in the Directorate for Mathematical and Physical Sciences,the Neural Systems Cluster's Organization and Modulation Programs in the Division of Integrative Organismal Systems, within the Directorate for Biological Sciences, the Theory Program in the Division of Materials Research in the Directorate for Mathematical and Physical Sciences and the Mathematical Biology program in the Division of Mathematical Sciences in the Directorate for Mathematical and Physical Sciences. This project is a first step towards the development of transformative theories and data analysis methods for studying the brain. In this project, the investigators will construct a theoretical framework at the interface between physics, engineering and biology that studies zero-one laws/phase transitions and related simplicities arising from having many degrees of freedom in the system. The resulting methods will be applied to both engineering examples and biological networks, particularly networks connecting neurons or brain regions. The investigators span a broad interdisciplinary spectrum ranging from theoretical physics and control theory to cell biology and neurobiology, and will engage in dissemination of the results and educating students and researchers across a broad spectrum in the subject matter of the proposal. In particular the researchers will pursue two neuroscience applications of the theoretical framework and tools developed in this proposal. The first will be directed to the analysis of experimentally determined neural circuits, and the second directed towards brain rhythms, with a focus on understanding the corresponding phase diagrams and the controllability of abnormal rhythms. They expect to recover the receptor field structure of the ganglion cells in the retina from such an analysis. These methods also will be applied to the mouse brain meso-circuit. Their analysis will go beyond graph-theoretic measures of the network and provide a bridge to activity measurements as well as to functional analysis of neuroanatomical circuitry, which is currently lacking. The investigators will also explore the consequences of the theoretical framework derived in this paper to networks of coupled, heterogeneous neural oscillators, with a distributed drive. The PIs will facilitate this impact by training researchers as well as holding meetings and workshops. The postdoc and the student will play an active role in organizing these two workshops. The PIs will publish preferentially with open access option, using funds from the grant. They will make all computer codes and secondary data analysis products available freely after publication of their results.

DESCRIPTION (provided by applicant): Ideally, as neuroscientists collect terabytes of image stacks, the data are automatically processed for open access and analysis. Yet, while several labs around the world are collecting data at unprecedented rates- up to terabytes per day-the computational technologies that facilitate streaming data-intensive computing remain absent. Also deploying data-intensive compute clusters is beyond the means and abilities of most experimental labs. This project will extend, develop, and deploy such technologies. To demonstrate these tools, we will utilize them in support of the ongoing mouse brain architecture (MBA) project, which already has amassed over 0.5 petabytes (PBs) of image data. The main computational challenges posed by these datasets are ones of scale. The tasks that follow remain relatively stereotyped across acquisition modalities. Until now, labs collecting data on this scale have been almost entirely isolated, left to reinvent the wheel for each of these problems. Moreover, the extant solutions are insufficient for a number of reasons: they often include numerous excel spreadsheets that rely on manual data entry, they lack scalable scientific database backends, and they run on ad hoc clusters not specifically designed for the computational tasks at hand. We aim to augment the current state of the art by implementing the following technological advancements into the MBA project pipeline: (1) Data Management will consist of a unified system that automatically captures metadata, launches processing pipelines, and provides quality control feedback in minutes instead of hours. (2) Data Processing tasks will run algorithms out-of-core, appropriate for their computational requirements, including registration, alignment, and semantic segmentation of cell bodies and processes. (3) Data Storage will automatically build databases for storing multimodal image data and extracted annotations learned from the machine vision algorithms. These databases will be spatially co-registered and stored on an optimized heterogeneous compute cluster. (4) Data Access will be automatically available to everyone-including all the image data and data derived products-via Web-services, including 3D viewing, downloading, and further processing. (5) Data Analytics will extend random graph models suitable for multiscale circuit graphs. RELEVANCE (See instructions): Nervous system disorders are responsible for approximately 30% of the total burden of illness in the United States. Whole brain neuroanatomy-available from massive neuroscientific image stacks-is widely believed to be a key missing link in our ability to prevent and treat such illnesses. Thus, this project aims to close this gap via the development and application of BIGDATA tools for management, storage, access, and analytics.

Cutting-edge light microscopy technology allows entire vertebrate brains to be digitized, resulting in data sets of unprecedented size and complexity. However there is a lack of adequate computational tools to visualize, manage, analyze, and disseminate these enormous data sets, and visual examination by expert human neuroanatomists remains the standard method to extract information from microscopic images. Software tools exist that can perform simple operations, but they are not able to adequately mimic the visual pattern recognition skills of an experienced neuroanatomist. This project aims to develop computational tools that mimic the analysis of an expert neuroanatomist, thus allowing for rich data analysis of whole-brain light microscopy data sets on a scale that has been previously intractable using human experts.

Machine vision algorithms will be developed and integrated into an an open source software toolbox (with associated whole brain image data) that will be made widely accessible for further development and refinement. Pattern recognition methodology will be applied to combine information about brain location (e.g., "where are we in the brain") with information about correspondence of brain structures in different species (e.g., "which areas of the brain of these species correspond"). Incorporating comparative neuroanatomical knowledge into pattern-recognition methodology is radically different from the "atlas morphing" approach currently used, and has the potential to transform the study of whole-brain neuroanatomy. The tools will help fill a gap in knowledge and skills (as contemporary neuroanatomists are increasingly less frequently being trained to study whole-brain microscopic anatomy), and postdoc and PhD students will be trained in the project.

? DESCRIPTION (provided by applicant): Presently there is no imaging technology capable of detecting neuronal activity in the entire human brain with millisecond and millimeter resolution. We propose to evaluate the possibility of developing a novel noninvasive method, sonoelectric tomography (SET), capable of directly imaging electrophysiological activity in the entire human brain with such resolution. In this method, conventional scalp electroencephalography (EEG) is used to measure the electrical activity. Each location of active tissue giving rise to the EEG signals is determined from the tagged US signature in the EEG signals. This information can be used to noninvasively construct a tomographic image of neuronal currents. In order to develop such a technique, we will evaluate three candidate mechanisms: (1) acousto-electric (AE) modulation of tissue resistivity, (2) mechanical vibration of the equivalent current dipole sources in active tissue, and (3) modulation of membrane properties. At present, it is still unknown which of these mechanisms can be used to implement the SET. We will first evaluate these mechanisms in rats in vivo. In Aim 1, we will apply a focused US to one region of the barrel cortex of the rat and test the sensitivity of the SET based on each mechanism. One barrel column will be activated by single whisker stimulation and the resulting local field potentials (LFPs) on the brain or scalp will be analyzed. The most viable US-encoding scheme will be determined from the US signatures in the LFPs. Aim 2 will be very similar to Aim 1, except a single linear US beam varying in US frequency along the beam will be applied to produce a one-dimensional image of neuronal activity. Aims 1 and 2 will establish the effective and safe mechanism for developing SET for human use. In Aim 3, the five digits of a hand will be stimulated with transcutaneous electrical stimulation to activate the finger areas in areas 3b. A linear US beam will be applied to each projection area in area 3b. The scalp EEG signals in area 3b will be analyzed for presence of EEG signals at the US frequency specific for each projection site. This will identify each active site. We will test to see if this method can identiy multiple active tissues. Alternatively, we will first identify the active tissues using a whole-hea MEG and/or EEG and then use the US to test the presence of activity at each predicted site. These tests will determine the feasibility and the best direction for developing a truly whole-brai US-based Activity mapping (USAmapping) technique.

Summary Progress from description to quantification is essential as a science matures. Yet numerical analysis of the elementary unit of brain circuitry?the individual neuron?continues to pose methodological challenges. Even the definition of a measurement yardstick (a metric) for the tree shape of a neuron remains an open research problem. Without such metrics, researchers cannot accurately classify neurons into cell types, an essential step toward understanding the circuit components and how they work together. Advanced methods from computational topology and geometry, which have only recently made their way from pure mathematics into data analysis, will be used to extract, characterize, and classify neuronal shapes in a way that elegantly incorporates the underlying dynamical electrophysiological properties. The first specific aim will apply new mathematical methods to define and compute metrics on the shapes of a wide variety of neurons. A computational topological analysis called ?persistence summaries? will be used to generate invariant representations of the neurons that can then be compared using different norms. An important strength of this method is that it works flexibly with arbitrary functions defined on the neurons, including purely structural ones (such as distance from the soma) or functions with electrophysiological meaning (such as electrotonic distance or propagation delays) and can therefore incorporate dynamics. A more advanced approach based on the Gromov-Hausdorff distance between metric spaces will be also explored. The metrics so generated will be used for classification and clustering, visualization of the space of neuronal shapes, and shape-based database search for neuronal reconstructions derived from light or electron microscopy. The second aim will use Morse theory to reconstruct individual neurons from light microscopic data, or skeletonize tracer injection data to summarize the structure of projection patterns. This approach retains shape information, which is lost when such data are characterized in a connectivity matrix. Further, these methods will be applied to construct consensus trees, which can be used as a summary of different reconstructions produced by different algorithms. The tools will be freely shared under a suitable open-source software license, and made available via plugins to widely used software platforms as well as web services to a community repository of neuronal morphologies. The team of researchers includes theorists, experimentalists, data scientists, and end users, all with extensive relevant experience. Apart from enabling the understanding of normal brain circuitry in terms of its component neurons, the proposed methods will also allow researchers to characterize changes in the shape of neurons in pathologically altered circuits, with applications to transgenic animal models of disease.

DATA CORE SUMMARY The effort to conduct a large-scale and comprehensive cell census within the mammalian brain will depend upon the systematic and careful handling of massive, multi-dimensional datasets. The proposed center will focus on generating a whole brain cell transcriptome atlas integrated with a forebrain cell anatomy atlas. Towards this end, the Data Core will support the other Cores and Research Segments by developing and implementing a common structural framework that can manage the data's inherent complexity. Specifically, the Data Core will map, manage, and analyze the single cell-level data generated across the project, including transcriptomes, dendritic/axonal morphology, and cell body location data. The common reference framework will be designed to permit web-based access and visualization of the primary data and facilitate efficient data transfer to the U24-funded data center (BCDC). Data Core lead Dr. Partha Mitra (CSHL) and co-lead, Dr. Aviv Regev (Broad Institute), will leverage existing frameworks designed to manage, analyze, and distribute high- resolution anatomical data (the Mouse Brain Architecture program at CSHL) and single cell transcriptome data (the Single Cell Portal at Broad). Specific activities of the Data Core will include developing statistically controlled high performance algorithms for biological knowledge discovery, data preprocessing and quality control, atlas mapping of individual brains onto a reference brain, and development and maintenance of a single unified data access portal.

PIs: Martin Deschênes, Yoav Freund, David Golomb, David Kleinfeld (lead), Fan Wang ! Core 2. Abstract Precision Histology This Core provides high-throughput tape transfer sectioning to cut near perfect stacks of slices of the whole brain, from olfactory bulb to spinal level C2. The tape transfer histology system has been in continuous operation in the Mitra laboratory for several years and will enable the registration of anatomical data from Research Projects 1, 2, and 3, into a common framework established by the Trainable Texture-based Digital Atlas. Control of orofacial actions is coordinated by distinct populations of brain stem premotor neurons that are arranged into relatively small clusters and limited to domains as small as 200 to 300 µm in extent. For many orofacial motor actions, premotor neuronal clusters are present at multiple levels of the brainstem and do not conform to the boundaries previously defined by available atlases, including the Paxinos atlases and the Allen Brain Common Coordinate Framework atlas. As part of our concerted effort to identify the functional organization of brainstem motor control, we require standardized processing of brain tissue. The established histology pipeline in the Mitra laboratory offers key advantages over alternative methods, in that tissue is retained in near perfect alignment during sectioning and is amenable to standard histological stains that demarcate brainstem landmarks used to align and annotate brains. Crucially, this approach precludes distortions from classical sectioning methods and enables efficient 3D reconstructions, cross-brain registration, and efficient incorporation of data from new brains into our recently developed Trainable Texture-based Digital Atlas. By efficiently multiplexing labeled neurons from transsynaptic tracing, in vivo recordings, and transgenic mice/cell type marker stains, the Precision Histology Core serves to support data compilation into a common anatomical framework.

Overview - Abstract Brainstem function is necessary for life-sustaining functions such as breathing and for survival functions, such as foraging for food. Individual motor actions are activated by specific brainstem cranial motor nuclei. The specificity of individual motor actions reflects the participation of motor nuclei in circuits within closed loops between sensors and muscle actuators. However, these loops are also nested and connect to feedback and feedforward pathways, which underlie coordination between orofacial motor actions. A key question for this proposal is how different actions are coordinated to form a rich repertoire of behaviors, such as rhythmic motions linked to breathing, and the orchestrated displacements of the head, nose, tongue, and vibrissae during exploration. We postulate that the best candidate interface for orofacial motor coordination are premotor and pre2motor neuron populations in the brainstem reticular formation: these neurons project to cranial motor nuclei, receive descending inputs from outside of the brainstem, and interconnected to each other. Our approach exploits and expands upon a broad spectrum of innovative experimental tools. These include state-of-the-art behavioral methods to study motor actions and their coordination into behaviors. From an experimental perspective, the underlying neuronal circuitry for each orofacial motor action may be accessed via transsynaptic transport starting at the muscle activators or associated sensors in the periphery. These studies will make use of molecular, genetic, and functional labeling methods to enable cell phenotyping and circuit tracing. These data will establish the Components, i.e., brainstem nuclei connectivity for all Research Projects. These studies are complemented by in vivo electrophysiology and optogenetics in order measure and perturb the signal flow during exploration and decision-making: these studies will establish orofacial ?Wiring Diagrams?. The sum of these techniques will permit us to elucidate the functions of intrinsic brainstem circuits and their modulation by descending pathways. Our data will be integrated in two ways. First we will begin development of computational models of the dynamics of active sensing by the orofacial motor plant and brainstem circuits. These will initially focus on the vibrissa system, starting with characterizations of mechanics and mechano-neuronal transformations of vibrissa movement and extending to exploration of brainstem circuits that drive vibrissa set-point and rhythmic whisking. Finally, vibrissa feedforward pathways will be computationally modeled to explore how sensory input affects vibrissa dynamics. Second, to record connectivity data that arises from our experimental tracing studies, we will construct an Trainable Texture-based Digital Atlas that utilizes machine learning to automate anatomical annotation of brainstem nuclei. The Atlas is designed to allow accurate 3D alignment of labeled neurons, even when labeled neurons reside in small sub-regions outside of well-defined brainstem nuclei, based on triangulation to Atlas landmark structures. Further, digitization of serially sectioned brain data sets allows 3D reconstruction and alignment of small brainstem subregions as well as the collation of this data from different brains into the same Atlas. Our proposed program on brainstem circuitry and dynamics will yield general lessons about the nature of neuronal computation. The analytic and anatomical tools developed for these studies will be made available through our data science core to the larger neuroscience community.

OVERALL SUMMARY   The  BRAIN  initiative  cell  census  network  calls  for  large-­scale,  comprehensive  approaches  to  define  the  composition of the mammalian brain at the cellular level and using an overall strategy that integrates multimodal  information  (morphology,  connectivity,  molecules  etc..)  within  a  Common  Coordinated  Framework  (CCF)  to  enable  distribution,  validation,  integration  and  use  of  the  atlas  by  the  community.  The  BICCN  challenge  is  enormous and remains a scientific problem requiring new discovery, continuous innovation in methods,  technologies and pipeline of analysis. Given the unparalleled cellular diversity of the mouse brain and the  need for an informed cell classification scheme, we propose here an ambitious project that addresses both the  need  for  scale  (coverage  of  millions  of  cells)  and  depth  of  analysis  of  each  cell  and,  further,  that  integrates  molecular and anatomical information. To address this challenge, we have assembled a collaborative group of  key knowledge leaders and innovators across various fields of neuroscience, genomics, and technology. First,  we  will  apply  transformative  new  droplet  scRNA  sequencing  technologies  and  next-­generation  computational  methods and data processing pipelines to compile a whole brain cell transcriptome atlas on a massive scale  (millions of single cells and nuclei collected brainwide). This effort will generate an unprecedented inventory of  cell type composition and distribution for the mouse brain within the CCF. Second, we will generate a forebrain  neuronal  atlas  that  will  integrate  detailed  molecular  information  (to  saturation)  of  anatomically  defined  populations with high-­resolution morphological and connectivity information to provide an in-­depth picture of a  core portion of the mammalian brain. We will also generate highly specific driver lines for precise marking of cell  types and to enable adaptive methods that refine cell sampling to achieve completeness. Finally, realizing the  need  for  innovation  in  technology  to  enable  work  that  is  made  difficult  because  it  requires  both  scale  and  precision,  we  will  devote  key  effort  to  develop  new  integrated  technological  platforms  that  combine  multiple  methods to relate neuronal connectivity with transcriptomes and cellular distribution at an unprecedented scale.  Our  Data  Core  will  integrate,  store,  and  manage  multi-­modal  datasets  and  provide  bioinformatics  and  computational expertise;? and our Administrative Core, will coordinate and oversee Center-­wide activities. Our  effort  is  unprecedented  for  scale  and  coverage,  and  it  relies  on  a  team  of  investigators  with  demonstrated  academic track records of innovation in technology and neurobiology, working in an environment that allows for  implementation  of  massive  pipelines  for  production  workflow.  This  will  guarantee  progressive  evolution  and  innovation of methods, experimental design and analysis to meet future challenges and succeed at generating  a comprehensive molecular and anatomical atlas of the mouse brain.