Part 3: CoCo analysis of replicate 1rhw simulations

It is claimed sometimes that, if you want to get good coverage of conformational space from MD, then it is better to run a series of independent shorter simulations, rather than just one long one. Here you will explore that assertion for this data set.

Exercise 4:

Use pyCoCo to analyse the first 25 ns chunks of five separate replicate simulations (which ever ones you like). Compare the variance and volume sampled for this ensemble of structures (a total of 2500 structures), with the values for the whole of replicate 1 you calculated earlier (also a total of 2500 structures).

Use the time remaining for this section of the workshop to investigate other aspects of CoCo and sampling. Possible questions could be:

1. How variable, in terms of the exploration of conformational space, are different replicate simulations?
2. If you keep adding more and more replicates to your data pool, does the volume sampled continually increase or is there a process of 'saturation'?
3. All the analyses done so far have been for sampling in 2 dimensions. What happens if you do the CoCo analysis in 3D or 4D - do the take-home messages stay the same?