Exploring relationships in gene expressions: a partial least squares approach.

摘要: Abstract Microarray technology has revolutionized the way gene functions are monitored. Analysis of microarray data is a fast growing research area that interfaces various disciplines such as biology, biochemistry, computer science, and statistics. While clustering classification techniques have been successfully employed to group genes based on similarity their expression patterns, much yet be learned about interrelationship levels among genes. We approach this problem with statistical technique called partial least squares capable modeling large number variables each relatively few observations. This property methodology appears attractive for application sets where simultaneous many collected at time points (or individuals). use it analyze publicly available sporulation budding yeast (Saccharomyces cerevisiae). investigate representative genes, one from temporal (based first induction) positively expressed show in case most variability was explained by only two regression terms all remaining Moreover, predicted fit very well average true over time. Finally, we compare biological largest coefficients those In cases, involved similar or related including negative relationships. method can identify established relationships; argue an exploratory tool identifying potential relationships requiring further investigation.