Examples of Current and Recent Projects

This section describes example projects that have involved Biomedical Informatics faculty, postdoctoral fellows, and graduate students. The projects are divided into four general areas:

Clinical Computing Within Yale New Haven Medical Center 

Faculty: Perry Miller, MD, PhD, Richard Shiffman, MD, Cynthia Brandt, MD, MPH, Prakash Nadkarni, MD, Mark Shifman, MD, PhD, Peter Gershkovich, MD, Seth Powsner, MD, Andrea Benin, MD, Allen Hsiao, MD, Ryan O’Connell, MD

In the early 1990s, Yale New Haven Hospital (YNHH) completed a 2-year process of installing CCSS, its hospital information system. CCSS provides mandatory physician order entry, results reporting, and other functions in support of clinical care. The medical center has also implemented an Ambulatory Care Information System (Logician) in many hospital and school clinics. YCMI faculty and fellows have collaborated closely on the planning and implementation of these systems, which provide the foundation for the ongoing development of an electronic medical record within the medical center. A more recent development, led by Prof. Shiffman, involves creating a community-wide health information exchange initially focused on the care of children with asthma. As a result, there are many opportunities for fellows to become involved in interesting clinical computing projects within both the Medical Center and the greater New Haven community.

Ongoing Collaboration with the VA Connecticut Healthcare System (VACHS) 

Faculty: Prakash Nadkarni, MD, Cynthia Brandt, MD, MPH, Luis Marenco, MD, Joseph Erdos, MD, PhD, Amy Justice, MD, PhD, Perry Miller, MD, PhD

Over the past decade, there have been a variety of collaborations between faculty at the YCMI and faculty at the VA Connecticut Healthcare System (VACHS), which is based in nearby West Haven. One project, directed by Prof. Joseph Erdos, Chief Information Officer of the VACHS, developed a pilot clinical data repository for the VACHS. It involved using a set of tools to extract patient data from the VA patient record system (written in the MUMPS programming language) and placing that data into a relational database to allow it to be queried and analyzed in a flexible fashion. One use of the data has been for provider profiling, where the model developed at Yale was adopted throughout the New England VA region. This project has also allowed us to explore a number of research issues involved in the design, implementation, and use of large clinical data repositories. A more recent collaboration between Profs. Cynthia Brandt (YCMI) and Amy Justice (VACHS) focuses on informatics research in clinical epidemiology and health services research. One goal is to use the VACHS electronic medical record system as a means of organizing and prioritizing care of HIV infection and important comorbid behaviors and conditions.

Trial/DB: A Web-accessible, Multi-disciplinary Database for Clinical Research 

Faculty: Prakash Nadkarni, MD, Cynthia Brandt, MD, MPH, Perry Miller, MD, PhD, Luis Marenco, MD, and many Yale faculty involved in clinical research

Trial/DB is a Web-accessible, multi-disciplinary database developed to store the data collected for clinical trials and clinical research projects. It was built in close collaboration with the Yale Cancer Center, and is also being used by Yale’s General Clinical Research Center and by many other Yale studies. It is also being used at several other medical centers. Research issues explored include: 1) how best to store clinical data using the Entity-Attribute-Value (EAV) format, 2) how best to allow users to query and analyze that data in a flexible, robust fashion, and 3) how to maintain the system's library of clinical terms in a well-organized and flexible framework. Prof. Nadkarni has also received support from the National Cancer Institute to use Trial/DB as the "special studies database" for its national Cancer Genetics Network (CGN), which includes eight academic centers nationwide where multi-institutional clinical studies are performed.

Clinical Applications of Mobile Computing Technologies 

Faculty: Richard Shiffman, MD

Personal Digital Assistants (PDAs) offer a lightweight, mobile platform that can be used at the point-of-care. Using several hand-held devices, we have developed software that can provide decision support and workflow support for clinicians in various specialties. Creation of effective human-computer interfaces is challenging. One related project evaluated the effectiveness of PDAs for support of national guidelines, with an initial focus on childhood asthma. Subsequent activities involve a collaboration with the University of Vermont to develop and deploy PDA-based decision support for smoking cessation in primary care.

Clinical Decision Support Systems and Computer-Based Clinical Practice Guidelines  

Faculty: Richard Shiffman, MD, Perry Miller, MD, PhD, Sandra Frawley, PhD, Fred Sayward, PhD, and many clinical faculty at Yale

A longstanding research activity has involved the development of programs which bring computer-based advice to the practicing clinician. A related emphasis has been on computer-based knowledge processing for a spectrum of clinical practice guidelines. This project explores the acquisition and representation of guideline knowledge. Previous activities have focused on defining methods for translation of guideline prose and flowcharts into rule sets, logical verification of rule sets, and simplification of comprehensive rule bases. We have developed a knowledge representation for clinical guidelines that has been standardized, and is being used in a growing number of settings worldwide. We have ongoing collaborations with national specialty societies, including the American Academy of Pediatrics, the American Academy of Family Physicians, and the American Academy of Otolaryngology.

Knowledge-Based Image Processing  

Faculty: James Duncan, PhD, Lawrence Staib, PhD, Hemant Tagare, PhD

The Yale Image Processing and Analysis Group, directed by Prof. James Duncan, is developing strategies to use information derived from biomedical images for 1) improved understanding of basic anatomical and physiological relationships in normal and disease states, 2) more accurate and reproducible clinical diagnosis, and 3) to help guide treatment. Projects have included 1) automated segmentation of the left ventricle of the heart from 4D cine Magnetic Resonance images (MRI) and of cortical and subcortical grey matter of the brain from static 3D MR images, 2) tracking and modeling nonrigid motion of the heart using shape-based strategies and 4D echocardiographic data, 3) indexing medical image databases according to information derived mathematically from the images, and 4) strategies for image-guided surgery and intervention, including MRI-guided epilepsy surgery, cone-beam CT-guided prostate radiotherapy, and targeted drug delivery for treating brain cancer.

Retrieval of Pathology Images Based on Image Content 

Faculty: Perry Miller, MD, PhD, James Duncan, PhD, Lawrence Staib, PhD, Hemant Tagare, PhD, John Sinard, MD, PhD, David Gelernter, PhD, Nicholas Carriero, PhD

This project developed a pathology image database system, PathMaster. The database contains images obtained by digitally imaging many cells, each indexed by computationally-derived descriptors. When a pathologist is confronted with a slide containing a cell whose diagnosis is uncertain, a digital image of the cell can be submitted to PathMaster. PathMaster automatically computes descriptors for the unknown cell and retrieves images of "similar" cells from its database. The images are passed back to the user along with their diagnoses, as a "visual differential diagnosis" to help the user classify the unknown cell. PathMaster was implemented using parallel computation as a pilot testbed for our Next Generation Internet project.

Neuroinformatics as Part of the National Human Brain Project 

Faculty: Perry Miller, MD, PhD, Gordon Shepherd, MD, PhD, Prakash Nadkarni, MD, Michael Hines, PhD, Luis Marenco, MD, Nian Liu, PhD, Chiquito Crasto, PhD, Kei Cheung, PhD, Avi Silberschatz, PhD, Drew McDermott, PhD

As part of the national Human Brain Project, we are developing informatics support of neuroscience research and computer-based modeling using the olfactory system as a pilot domain. Components of this project currently include: 1) ORDB, a database of information about olfactory receptors, 2) NeuronDB, a database of information about the compartmental properties of different neurons, and 3) ModelDB, a database of models of neurons and neuronal compartments. The project is providing a focus for developing a new approach for designing bioscience databases, the EAV/CR design (Entity-Attribute-Value with Classes and Relationships). This design facilitates the flexible storage and retrieval of complex bioscience data and of the biological relationships between those data items. In addition, we recently started participating in a national NIH-based collaboration to build a Neuroscience Informatics Framework (NIF) that will allow researchers to flexibly search and query a wide spectrum of Internet-based neuroscience resources.

Genetics, Genomics, & Yale’s Center of Excellence in Genome Science  

Faculty: Perry Miller, MD, PhD, Kenneth Kidd, PhD, Michael Snyder, PhD, Mark Gerstein, PhD, Sherman Weissman, MD, Prakash Nadkarni, MD, David Gelernter, PhD, Nicholas Carriero, PhD, Kei Cheung, PhD, Mark Shifman, MD, PhD

Over the past 15+ years, the YCMI has been involved in a number of projects in support of genomics and genetics. An early project involved exploring the use of parallel computation in sequence analysis, linkage analysis, and molecular dynamics, in collaboration with Prof. David Gelernter (Computer Science). Another project involved providing Internet-based informatics support for a collaborative Genome Center involving the Albert Einstein College of Medicine and Yale to map human chromosome 12. Other projects included: 1) developing a variety of databases that are used actively within the laboratory of Prof. Kenneth Kidd for cell lines, for phenotyping data, for primers, and for managing the laboratory analysis of a chromosomal region, and 2) working with Prof. Michael Snyder to build and refine a database for yeast gene expression data, as well as tools to help analyze that data. A major current focus involves working as part of Yale’s Center of Excellence in Genome Science (directed by Prof. Snyder), which is focusing on developing microarray technology and applying its in several biomedical settings, including the discovery of thousands of previously unknown transcriptional active regions (TARs) in the human genome.

Analysis of Genetic Regulatory Networks  

Faculty: Mark Gerstein, PhD, Michael Snyder, PhD, Sherman Weissman, MD

Prof. Gerstein and his collaborators have studied the structure of protein networks, both on a large-scale in terms of global statistics (e.g., the network diameter) and on a small-scale in terms of local network motifs. In particular, they have correlated network hubs with gene essentiality. Recently, they developed a number of tools to build and analyze networks derived from genes and also from literature citations. They have also investigated the dynamics of networks, i.e., how their topology changes over time. In particular, they have identified changing hubs and systematic patterns of connectivity rewiring in the yeast regulatory network.

Correlating Protein Abundance and Gene Expression 

Faculty: Mark Gerstein, PhD, Michael Snyder, PhD, Sherman Weissman, MD

Prof. Gerstein and his collaborators are involved in the correlation of mRNA expression and protein abundance. This project seeks to determine the overall correlation between these quantities across a number of experiments. They are also trying to identify particular outlier proteins that have anomalously high or low levels of abundance relative to their expression, and then to see if this discrepancy is related to certain proteomic or genomic features such as amino acid composition or the presence of upstream regulatory sites.

The Integrative YeastHub Project 

Faculty: Mark Gerstein, PhD, Kei Cheung, PhD, Michael Snyder, PhD, Martin Schultz, PhD

This research continues a longstanding set of collaborative activities by Prof. Gerstein that focus on the integrative analysis of genomic and proteomic data from many different perspectives. The focus of this specific project is on first collecting, annotating, and organizing a large set of diverse data involving yeast, and then performing integrative data mining in a variety of ways. This project is also exploring the latest Semantic Web technologies to facilitate integration and interoperation of diverse types of biological data from diverse types of Web-accessible data sources.

High Performance Computation in Biomedicine 

Faculty: Nicholas Carriero, PhD, Martin Schultz, PhD, Perry Miller, MD, PhD, Kei Cheung, PhD, Hongyu Zhao, PhD, Kenneth Williams, PhD, Joseph Chang, PhD, Mark Gerstein, PhD, Kevin White, PhD

This broadly focused project involves research into the use of High Performance Computation (HPC) in biomedicine. One subproject is exploring a “High Productivity/Low Maintenance” (HP/LM) approach to HPC in biomedicine, working collaboratively with a growing number of biomedical researchers. This project seeks to develop approaches and tools to make it easy for bioinformatics researchers to take advantage of HPC, while still maintaining their research codes (i.e., their computer programs) in a form that they can easily modify in an iterative fashion, (for example, in rapid application development environments such as R, Python, Perl, or Matlab). A second subproject involves exploring the use of parallel main memory database (PMMDB) technology that allows compute intensive computations involving large amounts of data (such as microarray data) to be performed in main memory rather then within a conventional relational database management system.

The GeneCube and Cruella Databases within the Department of Pathology  

Faculty: David Tuck, MD, Joseph Chang, PhD, Kei Cheung, PhD, Sherman Weissman, MD

Prof. David Tuck’s group is developing and refining GeneCube, an application developed in collaboration with Prof. Weissman’s lab. GeneCube currently provides a number of researchers at Yale and their collaborators with an integrated tool for analyzing data related to gene expression and proteomics research, including 1) hematopoietic stem cell development, and 2) function of mature blood cells of diverse lineages including neutrophils, monocytes, both T and B lymphocytes, as well as melanoma. Prof. Tuck has also developed Cruella, a tissue array database, for the Yale Department of Pathology. Prof. Tuck is also developing MDAService, a prototype Web services-based data analysis service. MDAService is designed to provide a set of complex analytic tools, available over the Web, that can be incorporated into a range of existing and planned applications, including analysis of data in GeneCube and Cruella.

Projects in Translational Informatics at the Intersection of Bioinformatics and Disease 

Faculty: Paul Lizardi, PhD, Josephine Hoh, PhD, Hongyu Zhao, PhD, Perry Miller, MD, PhD

Several recently inaugurated collaborations focus on translational bioinformatics. One project involves analyzing high-density, genome-wide Affymetrix SNP data to help identify the genes involved in disease, starting with Age-related Macular Degeneration (AMD), and using pathway data to help in this analysis. The second project involves using similar microarray data to analyze comparative genome hybridization (CGH) in patients with cancer, focusing initially on analyzing regional “copy number” changes.

The Yale Microarray Database  

Faculty: Perry Miller, MD, PhD, Kei Cheung, PhD, Kenneth Williams, PhD, Mark Gerstein, PhD, Kevin White, PhD

The Yale Microarray Database (YMD) is being developed as a university-wide database system that can support microarray research within Yale. This multi-tiered Oracle database has a Web interface, and will contain a variety of types of data, including 1) raw data consisting of digitally scanned images of hybridized arrays (stored on a file server and pointed to by the database), 2) the files produced by scanning these images using software such as Axon's GenePix, 3) information about each experiment, for example cell lines and experimental conditions, and 4) the results of various computational analyses, such as clustering experiments.

The Yale Protein Expression Database & Related Informatics Projects 

Faculty: Mark Shifman, MD, PhD, Kei Cheung, PhD, Kenneth Williams, PhD, Perry Miller, MD, PhD

The YCMI and Keck Biotechnology Center are collaborating to build the Yale Protein Expression Database (YPED) to help organize the processing of mass spectrometry proteomics data, which is being produced in increasingly large volumes by the Keck Center for many researchers at Yale and beyond. The database is designed to handle data from a variety of proteomics experiments including MALDI-MS based peptide/protein disease biomarker discovery, differential fluorescence 2D gel electrophoresis (DIGE), isotope-coded affinity tag (ICAT)/MS protein profiling, and multidimensional LC/MS analysis of tryptic digests of whole cell and partially purified protein extracts (MudPIT).