UMBRELLA SAMPILNG pull reference

GROMACS version:2022.5
GROMACS modification: Yes/No
I have performed the umbrella sampling simulations by pulling the protein away from the membrane.
pull = yes
pull_ncoords = 1 ; only one reaction coordinate
pull_ngroups = 2 ; two groups defining one reaction coordinate
pull_group1_name = PROTEIN
pull_group2_name = ECLIP
pull-pbc-ref-prev-step-com = yes
pull_coord1_type = umbrella ; harmonic potential
pull_coord1_geometry = distance ; simple distance increase
pull_coord1_dim = N N Y
pull_coord1_groups = 1 2
pull_coord1_start = yes ; define initial COM distance > 0
pull_coord1_rate = 0.005 ; 0.005 nm per ps
pull_coord1_k = 5000 ; kJ mol^-1 nm^-2

I got the PMF, but it did not show the potential well. To get the well, I want to pull it towards the membrane. Initially, my reference point was the protein COM and the membrane upper leaflet headgroup COM. When I pull the protein towards the membrane by setting
pull_coord1_rate = -0.005 ; 0.005 nm per ps

I get the error

Pull reference distance for coordinate 1 (-0.000041) needs to be non-negative

I suppose it’s because the protein is trying to cross the reference and, hence, getting a negative distance. Should I change the reference to the lower leaflet COM and pull it?
Will it cause an error in the WHAM calculation because changing the reference will make the distances inconsistent?

You should have used geometry=direction, not distance. My guess is that you can change to direction for the points close to zero and reuse all you current data with gmx wham.

Why are you using pull-pbc-ref-prev-step-com = yes? This should not be necessary.