Researchers continually upload data into public repositories at a rapid pace, yet utilize few common standards for annotation, making it close to impossible to compare or associate data across studies. To address this problem, we will develop a defined meta- data model and build an integrated system called Phenotype Discovery (PhD) that enables researchers to query and find genomic studies of interest in public repositories as well as upload new data into our database (sdGaP), in a standardized manner. A Query Interpreter (QI) will utilize text mining and natural language processing techniques to map free text into concepts in biomedical ontologies, allowing non-structured queries to be answered efficiently. In Phase I of the project, we will develop a proof-of-concept system that can retrospectively structure phenotypic descriptions in dbGaP, and will work with domain experts in pneumology to build use cases and evaluate the automated mappings. In Phase II of the project, we will extend the domain expertise to cardiology, hematology, and sleep disorders to build a more comprehensive system, expanding the phenotype annotation to transcriptome databases, and integrating a flexible automated genotype annotation tool for sdGaP. We will develop a user-friendly interface to prospectively assist researchers in uploading their data with standardized phenotypic annotations. We will provide the tool for free from our website and continuously improve its quality, based on user feedback and usage data.

Public Health Relevance

Phenotype Discovery (PhD) represents a novel, automated system to describe the characteristics of patients whose genetic information is available in public data repositories, without compromising their privacy. This initiative is greatly needed so that more researchers can make use of data collected from projects funded by public agencies. PhD uses new methodology for natural language processing and semantic integration to interpret the narrative text as well as variables and their values from studies in genomic databases. Standardized terminologies will be utilized to ensure that data can be analyzed across different studies.

National Institute of Health (NIH)
National Heart, Lung, and Blood Institute (NHLBI)
Exploratory/Developmental Cooperative Agreement Phase II (UH3)
Project #
Application #
Study Section
Special Emphasis Panel (ZHL1-CSR-K (M1))
Program Officer
Papanicolaou, George
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of California San Diego
Biostatistics & Other Math Sci
Schools of Arts and Sciences
La Jolla
United States
Zip Code
Doan, Son; Lin, Ko-Wei; Conway, Mike et al. (2014) PhenDisco: phenotype discovery system for the database of genotypes and phenotypes. J Am Med Inform Assoc 21:31-6
Doan, Son; Conway, Mike; Phuong, Tu Minh et al. (2014) Natural language processing in biomedicine: a unified system architecture overview. Methods Mol Biol 1168:275-94
Li, Pinghao; Wang, Shuang; Kim, Jihoon et al. (2013) DNA-COMPACT: DNA COMpression based on a pattern-aware contextual modeling technique. PLoS One 8:e80377
Lin, Ko-Wei; Tharp, Melissa; Conway, Mike et al. (2013) Feasibility of using Clinical Element Models (CEM) to standardize phenotype variables in the database of genotypes and phenotypes (dbGaP). PLoS One 8:e76384