The RNAseq examples in the section "Time course RNA-seq experiments of Drosophila melanogaster" are technical replicates. So, the reads can be summed by sumTechReps(). If I have biological replicates, how to sum the reads for the time course analysis (Natural cubic spline regression)? Thanks

If the samples are biological replicates, you don't sum the reads (i.e., keep them as separate samples). You can generate an MDS plot to see if there is any sample effect (i.e., separation by sample in one of the dimensions). If so, you may also consider incorporating the sample effect into the design.