Giving every gene the same epsilon causes inappropriate modeling of the x distribution, but integrating the variance into the model error results in large variances in the posterior. Normalizing the standard error as shown below alleviates these issues to some degree.
Giving every gene the same epsilon causes inappropriate modeling of the x distribution, but integrating the variance into the model error results in large variances in the posterior. Normalizing the standard error as shown below alleviates these issues to some degree.