0.001 were retained in the model. While the PIC showed the strongest relationship with the flow error-rate, we considered the replacement of PIC with simpler terms, such as the nucleotide flowed or pyrimidine versus purine, however this was detrimental to the model. Based on these choices, a simpler model was created and assessed against the full-model for significance using ANOVA. Some terms removed were statistically significant (x-pos, y-pos, machine), however were practically unimportant contributing only a very small amount to the modeled flow value. We emphasise that the purpose of this statistical model is not to test for significant factors, but to produce usable predictions.
