Identification of genes with bimodal expression. Muscle samples from 225 individuals were processed in groups that could be identified for distinct batch effects influencing the expression levels. Bimodality was assessed in groups of microarrays that were homogeneous for such bias. A variety of constraints were applied to ensure bimodal genes identified were genuine, including replication. Also illustrated is how bimodality may be artifactually (incorrectly) inferred due to heterogeneity of batch effects on expression levels.
Mason et al. BMC Genomics 2011 12:98 doi:10.1186/1471-2164-12-98