Design and Analysis Core

Since neuroscience-related data continue to increase rapidly in both volume and complexity, assessing its significance requires innovative and reproducible analytical tools and methods for efficient and reliable scientific inference. The mission of the Design and Analysis Core (DAC) is to support the COBRE projects, as well as the broad neuroscience community at Brown University, by providing expertise in principled study design, statistical modeling, machine learning, inference, and computation. The DAC functions by:

Planning: in direct collaboration with project leaders (PLs), setting examples and promoting highest standards in design and analysis; applying cutting-edge methodology in statistical analysis and developing novel methodology.

Supporting: managing synergistic relationship between different units of the computing infrastructure for smooth and robust data analysis, efficient computation and effective dissemination of both scientific results, models and novel analysis tools.

Enabling: training the PLs’ teams and local research community in cutting edge statistical methodology and computing tools.

DAC Goals

Our long-term goal is to grow into a sustainable resource that serves neuroscientists, psychologist and clinicians at Brown University, its affiliated hospitals and other Rhode Island institutions, by providing a broad range of statistical, machine learning, and high-performance computing services. To accomplish these goals, during the current five-years of support from the National Institute of General Medical Sciences (NIGMS), the DAC will focus on the following themes.

Provide project-specific, faculty-level collaboration to COBRE project PIs on design and analysis. Contemporary neuroscience research requires sophisticated data analysis and specialized expertise in study design beyond what standard software packages often offer. When involving collaborative efforts, it also requires that computational colleagues have a solid grasp of the scientific background along with a deep understanding of the characteristics of the data from particular technologies such as functional MRI and EEG. The DAC faculty bring a broad collection of expertise of statistical modeling, inference and machine learning to the COBRE projects, and will develop close collaboration with the PLs. With joint input from the Behavior and Neuroimaging Core (BNC), the DAC will support each project with experimental design, statistical analysis and machine learning. The DAC’s support includes identifying appropriate existing methods as well as adapting newly developed methods, often not available in the form of mature software, as needed by the PLs.

Develop novel statistical methodology and computing toolboxes for study design and comprehensive analysis of neuroimaging and behavioral data. The innovation in the COBRE projects will bring these projects to new frontiers as they progress and present challenges that demand novel development in statistical and machine learning methods. The demand from researchers doing innovative studies serves as the best motivation for novel statistical methodology development. The DAC faculty will develop new methods in response to the challenges faced in current research, apply their most recent development to the COBRE projects, and benefit from the follow-up and validation in the COBRE PLs' labs. These new methods not only directly benefit the projects, but are research products in their own right, and could have broad impacts for both the general neuroscience community and/or the statistics/computer science societies.

Coordinate and facilitate access to the Brown University central computing and information infrastructure and provide complementary support on data management. This is a shared aim with the BNC. The key objective is to enable the COBRE PLs to achieve efficient, reliable, and reproducible analysis of their research data; to maintain and manage their data and analysis process in a system that ensures continuity of research within their projects and lowers the boundary for cross-project transferring, as well as dissemination to the external scientific community. The DAC and BNC work in a modular approach to allow separate optimization in data acquisition and preprocessing at the BNC, and downstream analysis and inference at the DAC. The DAC ensures sustainable and reproducible analysis, and convenient reanalysis in case of changes in preprocessing choices, by maintaining and documenting the software environment, and using reproducible, dynamic document tools. The DAC coordinates the access to the high-performance computing servers, so that the COBRE projects take full advantage of the computing infrastructure at Brown.

Serve as a locus for training and exchange of ideas on computational neuroscience and extend the statistical and computational support to the growing community of neuroscience researchers in Rhode Island. The DAC aims to form/strengthen synergistic relationship with the Brown Data Science Initiative, the Biostatistics Core and the Biomedical Informatics Core of the Rhode Island Advance Clinical and Translational Research center, and the recently funded COBRE Center for Computational Biology of Human Disease, and to form a coalition in training and fostering strong and sustained collaboration between data scientists (statisticians, computer scientists and applied mathematicians) and neuroscientists, psychologists and clinicians. The core will build a platform to disseminate the methods and tools (new methodology developed in-house or adapted from recent publications) that promote a broader impact of the COBRE research to the wider community. This also enables local researchers to learn and identify potential collaboration projects.