AT A GLANCE Received his M.D. from Johns Hopkins University. Completed his residency at Massachusetts General Hospital in Boston.
Associate professor of medicine, University of Colorado at Boulder. Research specializes in pulmonary medicine, and in particular gene expression patterns in pulmonary disease.
Outside interests include spending time with his wife of 20 years and four children, and skiing.
QWhat role do microarrays play in research at the University of Colorado?
AWe have over 80 Principal Investigators who’ve used the facility. Most come from the school of medicine, doing pharmacological research, and even surgery research.
[My research group] had the lead article in the March issue of Circulation, which showed that gene expression patterns were different in the familial form of pulmonary hypertension compared to the sporadic form.
QHow is the microarray facility set up within the university?
ACustomers consult with us to define the best way for them to design and run experiments. Then they come to the lab and run their experiments with the support people in the core. The wet lab reagents are allocated in single-use packages. They do the hybridization and scanning. Or if they are not skilled in molecular biology, we can provide all of the labeling, hybridization, and scanning. Then we go over the data analysis with them. We also have recruited a very large bioinformatics group.
QWhat types of microarrays do you use and in what combination?
AWe run both Affymetrix arrays and cDNA arrays made in-house. We run about 1,200 Affymetrix arrays a year, and used to be the largest academic Affymetrix core facility in the country. In addition to the University of Colorado, we service some small biotech firms in Boulder and the University of Wyoming. But we may have saturated our market.
QHave you developed any special protocols for optimization of array performance?
ASpotting in Denver, with a humidity of 10 percent is phenomenally difficult. The chambers all have to be humidified, a task which Brian Soriano, manager of the spotting facility, handles.
QWhat kind of arrayer do you use?
AA Cartesian 7500 robot. We also have an Axon GenePix 4000 scanner, and a Qiagen 8000 robot for plasmid and oligo preps. We are buying a GenePix 5000 scannner and new-generation Cartesian arrayers.
QWhat software do you use to analyze microarray data?
AWe have the Affymetrix micro- array suite, and are one of the only cores that have their Laboratory Information Management Software (LIMS) system along with the expression data mining tools. We also have 40 seat licenses, along with 10 floating licenses for Silicon Genetics’ GeneSpring software; and we have the BioDiscovery package, along with Axon’s GenePix Pro software for our cDNA facility. The bioinformatics group also develops homegrown software, and runs a fee-for-service facility for data analysis.
GeneSpring has been the most user-friendly. We have so many site licenses that the people from Silicon Genetics come out quarterly to give us training updates. With Affymetrix we have had no problems in terms of accuracy, but their clustering algorithms, their nearest neighbor analysis, and graphical representation have not been as user-friendly as GeneSpring.
QWhat is the biggest challenge you face in working with microarrays?
AData analysis and validation. Most journals want you to validate your findings through independent means. As part of our facility we have the PerkinElmer 7700, which we can use to validate genes of interest through quantitative PCR.
QHow do you tackle these challenges?
AWe have not only hired the four bioinformatics folks, but also have a tremendous amount of support from the commercially available software.
QIf you could make out a wish list for microarray technology advances or improvements over the next couple of years, what do you most want or need?
AFirst, everybody needs a standardized labeling or hybridization protocol that can be used so we can individually validate things. The quality of the arrays and clone sets is number two. Commercial arrays and in-house arrays need to be made with high-quality clones and preparations so the data can be shared.