When we use mixed effects modeling to analyze time course data where there is a a large number of time points, it is very likely that the mode does not converge. This is because there are too many parameters to estimate and there are not enough data points to estimate so many parameters in the data.
R cannot help us solve this problem because it assumes unstructured variance-covariance structure and does not allow the specification of different variance-covariance structures. If we stick to R, we would have to slightly modify the random component in the model. Here we use the SAS program to analyze the data. SAS allows us to chance variance-covariance structures so that we can reduce the number of parameters in statitical etimation and then solve the convergence issue without having to change the model.