Institute Awards Seed Funding to Nine Research Teams

In order to support its research strategy to invest in the future, the Hariri Institute for Computing selectively seed funds new collaborations that cross typical disciplinary boundaries as well as support ambitious research and forward-looking education initiatives.  Current and previously funded projects represent a broad array of the exciting new computing-related and data-driven research happening across the BU community.

The Institute is pleased to announce the recipients of this round of Hariri Research Award funding. Nine teams, working on a spectrum of projects from record linkage to refugee assistance and human neuroimaging research, will be supported through Institute seed funding. Many of the teams will also be working with software engineers at the Institute’s Software & Application Innovation Lab to realize the results of their research.

The recently awarded projects awarded are:

IKEA: Product, Pricing, and Exchange Rate Pass-through
PI: Marianne Baxter, Economics, CAS
Co-PI: Margrit Betke, Computer Science, CAS
The project seeks to use detailed catalog information on products and page layout to advance an understanding of the decision-making process of IKEA – the world’s largest furniture retailer, with stores in 41 countries and annual revenue estimated at $24 billion. The project will achieve this aim through two components: (1) Construct a new data set of IKEA product specifications and prices from catalogs from 1988-2004 in seven countries: US; Canada; France, Germany, Italy, UK, and Sweden for the purpose of understanding IKEA’s process of product creation and international coordination of price setting. These data will be linked with an existing data set covering 2005-2014, providing a lens to IKEA’s response to the introduction of the Euro and the global Great Recession; and (2) Analyze the visual layout of the IKEA catalogs to detect cultural/national differences in visual and text-based product presentation.

The Relationship between Cult and the Natural Landscape in Ancient Greece
PI: Andrea Berlin, Archaeology, CAS
The project aims to develop a model that interrogates the visual and spatial relationships between Greek sanctuaries and the natural landscape. The major innovation of project is the plan to collate, preserve, and present data in an online, interactive platform so that future users can perform their own analytical studies. The platform will allow a user to access, for any given place, all of the attributes and visuals collected as well as then be able to manipulate and study them. The results of the analyses, including visualscapes and topographic and cultural observations, will be made available as well.

eMap: Online Mapping Electron Transfer Channels in Proteins
PI: Ksenia Bravaya, Chemistry, CAS
This project is developing a computational platform for automated pre-screening of protein X-ray structure for efficient ET channels from user-defined source residue or co-factor to surface-exposed amino acids and its implementation as a robust web application. The developed web-based application, eMap, will enable prediction of electron transfer channels based on the protein crystal structure using graph theory and the distance-dependent electron tunneling penalty functions. eMap will be freely available for students, faculty, and researchers from Boston University and from other educational institutions.

Nielsen Retail Scanner Data Database
PI: Adam Guren, Economics, CAS
This project seeks to create a scalable structural database that allows for relational queries to make the Nielsen data more easily used for research. This project will have broader impacts across Boston University as many faculty across the university have expressed interest in extracting from this rich data resource. Having a database for the Nielsen Scanner Data would be of great help to a number of research projects and would more broadly help spur research projects at BU interested in consumer and firm behavior.

Social Networks at Microscale
PI: Kirill Korolev, Physics, CAS
Co-PIs: Daniel Segre, Biology, Bioinformatics, and Biomedical Engineering (CAS & ENG)
The project aims to develop a computational tool that can use microbial genomes to predict microbial interactions and ultimately ecosystem dynamics. Beyond prediction, this computational tool will be used to develop design principles for artificial communities and infer microbial interactions in medical and environmental microbiomes. The latter constitutes a major challenge faced by the microbiome research today as interactions are inferred from patterns of co-occurrence in cross-sectional data or temporal correlations in longitudinal data. The computational tool will constrain interaction types using genome-scale metabolic models and remove network motifs inconsistent with the need to co-localize in space. As a result, it will considerably improve the power of sequencing surveys to identify interactions.

Urban Refuge: Putting Aid on the Map
PI: Noora Lori, International Relations, Pardee School of Global Studies
The project seeks to create the backend of the Urban Refuge, beta-test and launch the product in Amman in 2017. Urban Refuge is a smartphone application providing refugees with new tools for navigating insecurity and managing crises. Furthermore, Urban Refuge enhances the digital infrastructure of cities, and the data collected from the app will be used in faculty and graduate student research to revise the way aid is allocated and distributed in cities, particularly among displaced populations. This phase of developing the application will yield the following outcomes: (A) Data can be collected from app for faculty and graduate student research in determining spatial concentration of aid; (B) Revise the way aid is allocated and distributed in cities as part of a larger mapping project.

Toward a Cloud-based ‘Reproducibility Engine’ for Human NeuroImaging Research
PI: David Somers, Psychological & Brain Sciences, CAS
Co-PIs: David Osher, Psychological & Brain Sciences, CAS
The project proposes the establishment of a reproducibility engine for neuroimagers, employing the power of the Massachusetts Open Cloud (MOC), which will empower researchers to easily and quickly analyze hundreds of subjects and either replicate, or fail to replicate their laboratory-based findings. The scientific scope of the reproducibility engine is to examine functional connectivity of networks in the human brain. The plan will enable users to upload brain activation results from their (small N) task-based scans and examine the functional connectivity of these regions using the HCP dataset. Furthermore, this project will serve as a springboard in the development of other cloud-based analytical tools for reproducibility and large-scale data mining with the HCP. The development of these tools should also serve the growing neuroimaging community and the BU Cognitive Neuroimaging Center that will open in the CILSE building this spring.

Statistical Physics of Nonstationary Long-Range Correlated Neural Time Series
PI: Gene Stanley, Physics, Chemistry, Biomedical Engineering and Physiology (CAS, ENG & MED)
Co-PIs: Marc Howard, Psychological & Brain Sciences, CAS; Howard Eichenbaum, Psychological & Brain Sciences, CAS
This project aims to study neural time series in a wide range of animal models and brain regions using sophisticated methods from statistical physics. It will use an extensive existing database of single-unit recordings and collect a new dataset using two-photon microscopy to measure neural activation over long time scales – a technique which is not yet in wide use. This research will aid in understanding the computational function of the brain, one of the most important open problems in science while solving this problem will have immense technological implications and shed light on disciplines that rely on human intelligence, including education, medicine, and economics.

Machine Learning-Driven Quantification of Pathological Fibrosis for Prognostic Relevance
PI: Vitaya Kolachalama, Medicine, MED
Co-PIs: Katya Ravid, Medicine, MED; Vipul Chitalia, Medicine, MED; David Salant, Medicine, MED
The project proposes to use artificial intelligence (AI) methods to derive quantitative information from the kidney biopsy images that can guide the development of effective treatments.  The burden of chronic kidney disease is enormous both for patients and the healthcare system, and yet we have few treatments to prevent patients from developing kidney failure and the need for dialysis or a kidney transplant. Currently, clinicians rely on a procedure called kidney biopsy (a tiny piece of kidney tissue for microscopic analysis) that is examined in a subjective fashion to identify diseases that might be amenable to treatment. Successful completion of the project can add an objective dimension to kidney biopsy analysis, with broad appeal to the entire renal community.

For technical/research-related questions about Hariri Research Awards, please contact Andrei Lapets, Director of Research Development, at lapets@bu.edu.

For administrative questions (process, budget, etc.) about the Hariri Research Awards program, please contact Linda Grosser, Director, Program and Project Management, at lgrosser@bu.edu.