Deriving clusters of genes by different clustering techniques or finding the statistically significant variations among genes are conventional approaches to study microarray expression data. Nowadays in vitro experiments are being considered to make applications of genetical genomics more widespread in non-model species. Different bioinformatics tools are being used to investigate genetic pathways in the form of correlation based networks. In this study, a comparison was made between in vivo and in vitro gene expression data by using two software: BioLayout and GeneNet. From ten mice, five mice with the wild-type allele and five mice with the gene knock out (KO) for the gene SOCS2, a total of twenty samples were taken: five fresh samples from wildtype mice, five fresh samples from KO mice, five cultured samples from wildtype mice and five cultured samples from KO mice. After obtaining differentially expressed genes from microarray cDNA experiments, network analysis was done using the software BioLayout and GeneNet to make correlation and partial-correlation based networks. The resulting networks, or clusters derived from the networks, were subsequently analyzed for gene set enrichment analysis (GSEA) using the tool DAVID. The results from the GSEA were used to compare all the clusters and networks between the fresh and cultured samples to test for functional overlap. The GSEA results were also used to compare the clusters from BioLayout with the networks from GeneNet to compare overlap between these tools using the same data. When functional enrichment analysis and comparisons were made between the fresh and cultured data set after getting the networks and clusters from BioLayout and GeneNet, only a few functional categories were found in common. This suggested that in vitro samples are unable to give the same biological information as in vivo samples for this particular gene KO. Also the two different network tools showed only limited overlap, suggesting that the correlation based networks from BioLayout show a different type of relationship among the genes than the partial correlations from GeneNet.
Therefore, the use of different network tools can be recommended to visualize and explore the regulatory pathways among genes.