Skip to main content
Fig. 2 | BMC Genomics

Fig. 2

From: Epistasis analysis of microRNAs on pathological stages in colon cancer based on an Empirical Bayesian Elastic Net method

Fig. 2

The overall workflow of our epistasis analysis based on EBEN. Step 1: Run the simplified model including solely main effect, which means only the significant features from this step are included in the main effect part of the model at the final step, Step 2: Obtain the corrected phenotype y through removing main effect of significant features from the original phenotype y, Step 3: The newly corrected y is used to infer epistasis. Only the features with significant epistatic effect can be included in the model in the next step. Step 4: Run the full model that includes all the features with significant features with main effects from the first step and significant features with epistatic effect identified in third step. EBEN is used as the parameter estimation method in Steps 1, 3 and 4. Here, y denotes the trait phenotype, X represents the miRNA expression, β m and β e represents for the main effect and epistatic effect separately, μ represents for the phenotype mean and e represents for the standard error

Back to article page