Omparison test showed that even PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/27889546 after removing one of the datasets from analysis, thus reducing the number of genes in the cluster, the functional enrichment remained highly similar. This shows that the co-expression of genes in this cluster is not a random effect, but the correlation ranks from DuF2 dataset has a large effect on the clustering process. This variation in the number of genes in the cluster indication of the effect of technical variabilities in high throughput results and shows the importance of validating this cluster on additional datasets from multiple pig populations. To our knowledge, this study is one of the first attempt in porcine androstenone research community to understand population similarity in gene expression patterns based on co-expression networks. With this study, we aim to provide a baseline co-expression cluster focusing on population similarity in gene expression patterns. This cluster can further be expanded or challenged based on analysis results from other porcine populations or breeds with similar androstenone phenotypes. In order to understand the breed differences in androstenone metabolism, as a first step it is crucial to know the breed similarities in androstenone metabolism. By validating the existence of majority of the genes in this cluster in various pig breeds it would be possible to eliminate the breed specific genes from the cluster and obtain a cluster of genes common for all the pig population. Once we obtain such a common cluster, it would be possible to Obatoclax rank the genes in the cluster based on either their correlation coefficients/jointSahadevan et al. BMC Genetics (2015) 16:Page 16 ofCDF to other genes in the cluster or based on expression values from high-throughput results. In the final step the ranking of these genes can be used as starting point for screening the animals. To conclude, we propose our co-expression cluster as one of the first attempt towards understanding gene expression similarities in hepatic androstenone metabolism. It is necessary to further validate this cluster in additional porcine populations (breeds) and to understand the potential roles PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/28552557 of member genes in androstenone metabolism. For this purpose large scale experiments including data from multiple porcine population combining data from genomic, proteomic and metabolomic experiments are necessary.(SCAI), Schloss Birlinghoven, 53754 Sankt Augustin, Germany. 3 Department of Animal Science, Faculty of Agriculture, Erciyes University, Kayseri, Turkey. 4 Department of Animal Production and Technology, Bogor Agricultural University, Bogor, Indonesia. Received: 18 July 2014 Accepted: 18 DecemberAdditional filesAdditional file 1: Data density plots for DuF2, Duroc and Landrace datasets. Additional file containing data density plots. Additional file 2: Pictorial representation of analysis workflow used. Legend: White parallelograms with grey outline: Input/output data and results. White cylinders with red outline: data from external databases. Rectangles with light blue shades: various tools and analysis processes used in this workflow. Additional file 3: LA GO and KEGG enrichment table. Additional table containing GO and KEGG enrichment analysis results for LA clusters. Additional file 4: HA GO and KEGG enrichment table. Additional table containing GO and KEGG enrichment analysis results for HA clusters. Additional file 5: LA cluster 2 network file. Cytoscape (.xgmml) file for LA cluster 2. Each edge in the netw.