Sample schedules

Sampling

Choosing an appropriate selection of points to record for an NUS experiment is very important, however, the understanding of this topic is largely based on heuristic studies. There are two things to consider:

• sampling percentage

• sampling pattern 

A variety of different patterns have been proposed in the literature. Current understanding suggests that biasing samples towards the earliest time-points (i.e. those with greatest signal-to-noise) and avoiding large gaps is important. Two popular sampling patterns are exponentially weighted, with the decay of samples related to the R₂ rates of the protein for the nuclei observed in the indirect dimensions, and Poisson-gap sampling where gaps between points are based on a Poisson distribution to minimise large gaps/'clumping' of points. Poisson-gap sampling is typically weighted by a sinusoid biasing points to earlier evolution times.

It is often helpful to take a full-sampled experiment and artificially undersample it to test the effects of different schedules before recording an NUS experiment. This can be done with Cambridge CS software.