AT A GLANCE: Received his PhD in molecular biology from UCLA in 1989. Worked with Michael Grunstein to examine the function of histone H4 in S. cerevisiae. Post-doctoral research at CalTech with Paul Sternberg, exploring gene function in C. elegans development. Set up microarray facility at SmithKline Beecham, now GlaxoSmithKline, in King of Prussia, Pa. Joined Bristol-Myers Squibb’s transcriptional profiling group in 1999.
QWhat role do Microarrays play in research at Bristol-Myers Squibb?
AMicroarrays are integrated into many of the activities of our pharmaceutical research institute. Arrays are used for identifying and validating drug targets, monitoring efficacy of compounds, identifying biomarkers, and exploring the toxicological aspect of compounds.
QWhat types of microarrays do you use?
AWe manufacture a variety of arrays in-house, in addition to purchasing commercial arrays. We choose a platform based on the requirements of a given experiment.
QWhat is the biggest challenge you face in working with microarrays?
AThe biggest challenge is that user effects are still a large component of microarray performance — i.e., many current protocols will have varying success depending on the user. Setting up a hybridization and the subsequent washing steps have a great impact on the final image produced. We would like to streamline the process to reduce these user effects and make it even more robust and higher throughput.
QHow do you tackle this challenge?
AWe are evaluating automated slide processors and designing new hybridization setups with the specific aim of increasing throughput, while also minimizing the variabililty in hybridization setup and washing steps. We are also working with our bioinformatics group to streamline data entry and project tracking.
QWhat techniques do you use to analyze micro- array data?
AWe use a number of microarray data analysis techniques, from simple outlier detection to descriminant predictors, a small set of genes whose expression patterns can discriminate between conditions such as disease type or compound treatment. What we find more important than the way data is analyzed is ensuring that numbers going in are of high quality. We typically use replicates of all samples and then statistical analysis to calculate proper values while assigning error values to these numbers. After this, we choose the most appropriate analysis for the questions we are trying to answer.
QHow is your microarray unit organized within the company?
AAs the manufacturing of microarrays must be tightly controlled, all arrays for in-house use are produced by one lab. Use of microarrays is open to all researchers as needed.
The transcriptional profiling group at Bristol-Myers Squibb provides training for use of a number of types of arrays and is also involved in many collaborations throughout the company. With increased success, there has been greater interest from a number of internal research areas.
QIf you could make out a wish list for microarray technology advances or improvements you would like to see materialize over the next couple of years, what do you most want or need?
AI would like to see lower cost and improved throughput for microarrays. This would permit the use of more replicates, which would improve data quality immensely. The greater availability would also permit such experiments as profiling large numbers (think thousands) of compounds in appropriate tissues and cell lines to better understand mechanisms of action and toxicity.