Hi, I have a time-course microarray data. 7 subjects, each one has several time points (Day0, Day10, Day20, and Day30). And the trait data also have the same time point for every subject. 

I wonder if make sense to pool all time points data together to construct the network and then correlate with trait? Or separately construct a network for each time point? 

Thanks.

It depends on what your research aims/questions are. However, since you only seem to have a total of 7*4=28 samples, I would construct one network from all samples. If you find that the largest driver of variability is uninteresting, adjust the data for it and construct a new network.