[gmx-users] gmx cluster for loop motions

[James Starlight]

Dear Gromacs users!

I am using gmx cluster (default clustering method, decreasing cut-off
to 0.025 -fit option) to investigate motions of the loop region of
water-soluble protein observed in 100 ns of MD simulation with the aim
to obtaine several snapshots corresponded to different clusters

In my task there are two proteins with similar 3D structures:
1) for one protein (where the loop sampled more distinct states ~ 3-4
A compared to starting model) the method of clasterisation works fine
giving 5 distinct clusters (e.g. the same was observed on PCA
projections calculated for the same region)

2) however for the second protein where the loop sampled just a few
states the gmx cluster with the same options has crushed. I tried to
varry cut-offs but it did not work. Only when the selection for "all
protein" was provided instead of "only atoms for the loop" it produced
reasonable results. Did I something wrong in my analysis?

Thanks you in advance!