A challenging task in time series microarray data analysis is to identify co-expressed groups of genes from a large input space. The overall objective of this study is to obtain knowledge about the most important genes and clusters related to production and growth rate in a real-world microarray data analysis task. Various measures are engaged to evaluate the importance of each gene and to group genes based on their correlation with the output and each other. Some strategies for grouping and selecting genes are integrated resulting in several models tested for real biological data. All proposed models are tested on a real microarray data analysis problem and the results obtained are throughtly presented as well as interpreted from a biological perspective.