Few basic questions regarding building MSM using pyemma.

[udnenhsirK]

I want to understand the how the thermodynamics and kinetics has changed due to ligand binding. So we have performed several short (1 microsecond) MD simulation with and without ligand.
We are using several CA-CA distance as feature.
The problems we are facing

1. If we include too many contacts the timescales are not converging.
2. When comparing two systems for example apo and ligand bound, should we perform tica for apo and bound trajectories separately or it can be performed on combined trajectories ( not concatenated by as a list of trajectories for apo and bound together) ?
3. In case of BPTI tutorial, when doing TICA, dim=2 was chosen. Is it just for the tutorial or we can choose only 2dimensions to perform K-means clustering.
4. How do we choose the number of macrostates or metastable states? It is recommended that if there is n slow process then number of metastable state is n+1. How do we define this slow processes? Can we define it by certain timecales? For example we got some timescales after building MSM. Now we find there are 3 processes that are slower than 300ns. Can we choose 4 states for coarse graining?
5. What are the necessary tests that I need to perform to validate the MSM other than CK-test.

I am new to this field so I apologize for the very basic questions.

[clonker]

Hi, may I direct you to the following notebooks in which all your questions are addressed with example data: https://github.com/markovmodel/pyemma-workshop/tree/deeptime/notebooks :)

Hope this helps