 |
The Department of Biomedical Informatics, under the direction of Hai Hu, Ph.D., shoulders two responsibilities in the Institute; one is to provide informatics and analytical support to the institute’s daily operations, and the other is to conduct biomedical informatics research. We are equipped with multi-processor computer systems in both Windows and Unix for development, testing, and production. The currently hosted applications include, a Laboratory Information Management System (LIMS), a data warehouse, and a variety of commercial software and in-house developed application tools. |
LIMS
WRI currently deploys the Laboratory Workflow System (LWS) from Cimarron Software (Salt Lake City, UT) to track genomic and proteomics experiments involving microarray, genotyping, DNA sequencing, and 2-Dimensional Difference in Gel-Electrophoresis/Mass Spectrometry (2D-DIGE/MS). In addition, we co-developed, with Cimarron, a Clinical LWS (CLWS) system to track operations at the clinical end. The CLWS connects clinical data to wet-lab based operations. We are conscientious of new technologies that appear on the horizon constantly, and we are making efforts to enhance our LIMS capabilities to support the new technologies being deployed at the Institute.
Data warehouse and data models
WRI generates a large amount of genomic and proteomic experimental data in addition to the clinical data we collect. A hybrid data warehouse has been envisioned to integrate and federate internal data from all the platforms in addition to external public data for integrative biomedical informatics research. The current development, in collaboration with InforSense Ltd. (London, UK), is Oracle based. A production version of this data warehouse, focusing on clinical data with an On-Line Analytical Processing tool (OLAP), has been released for use by WRI and its partners.
A questionnaire-based data warehouse focusing on breast disease studies has been developed and deployed. To improve our data warehouse for better handling of temporal data and easy expansion to support future clinical/translational projects, we started to develop a generalized patient-centric, object-oriented data model containing disease-independent (such as demographics) and disease-specific objects, including temporal and non-temporal objects. This new data model will allow for the easy dissection of study participants’ clinical and molecular data. In addition, when a new disease type is studied, most of the new data can be stored in existing data objects and only a few new disease-specific objects need to be developed, thus allowing easy adaptation of the objects to new disease studies. The same principles apply to the design of new questionnaires using such a data object structure. This new phase of the data warehouse is now in beta testing and should be deployed by the end of the year 2007.
QA/QC
At WRI we hold the opinion that the quality of the data lays the foundation of the subsequent data analysis and experiments. We are determined to strive for the highest data quality. To obtain high-quality data many of our QA/QC procedures have been embedded into our SOPs. For example, the clinical QA program consists of: 1) a visual inspection of missing values and obvious inconsistencies, 2) double data-entry to reduce data entry errors, and 3) a computer program that deploys established QA metrics to check for data integrity across the whole questionnaire. A microarray QA program can identify outlier slides from a population of similar slides. Additional QA programs have been developed or are in the process of being developed to ensure the quality of the data generated at WRI.
Research projects and new application developments
Our research projects span across clinical, genomic, and proteomic areas. We not only help lab scientists design experiments, perform QA, and conduct data analysis, but also carry out biomedical informatics research ourselves. Since biomedical informatics research is relatively new not all the studies can be conducted using existing applications, so novel tools need to be developed. The following is a list of a few active research and development projects:
- A study on pathology diagnosis co-occurrence and the development of novel tools to visualize study results
- Use of Gene Ontology-derived similarity measures for quantitative characterization of direct and indirect protein interactions within human regulatory pathways
- Neural network modeling using clinical and molecular study data for patient classifications
- Development of a new LC-MS data analysis method
- Development of a Quality Assurance Issue Tracking (QAIT) system