• View
• Edit
• History
• Print
• Backlinks

Piggybacking

From our Translocation Stages protocol (11 rounds of energy minimization, 20ps of heating, and 3ns of equilibration) we were able to generate fairly stable trajectories for four of the five translocation stages. The run stabilities (defined as trajectories that ran for their entire specified time) were 100% for 5JUO, 100% for 5JUS, 81% for 5JUT, and 93% for 5JUU. However, only 36% of 5JUP runs went to completion. This indicated the need to modify our protocol.

Local Approach

We first considered trying to address areas of local high energy to see if we could improve run stability through an adjustment of the onion shell for a handful of residues. Theoretically, this approach is preferable because it does not allow the entire subsystem to wander away from the cryo-EM structure -- our only tie to biological relevance. This was implemented by running energy minimization on the structures of 5JUP written right before the run crashed. The energy minimization output includes the identity and energy of the highest-energy atom in the system. We then looked at repeat offenders across several trajectories and decided on the following modifications to the onion shell:

   379-444 → 379-440
   451-479 → 451-477
   269-312 → 269-275, 277-312

Unfortunately, in testing of this protocol 13 of the 21 runs crashed, including runs that did not crash with the old onion shell. We therefore decided to test a different approach, as specified below.

Global Approach

We next considered adding an extra step before neutral dynamics to allow for the structure at the end of equilibration to evolve (with only the onion shell restrained) before being fixed as the reference structure for neutral dynamics. Because we are inheriting the reference structures sequentially from equilibration to this additional step to dynamics, we call it a "piggyback" step.

Several variations of this were tested, including the addition of multiple (3 and 2) piggyback steps, and one piggyback step of 1ns and 500ps. The 3x1ns piggyback protocol was tested for GAU, GCU, GGU, and GUU. We found that GUU was the most dissimilar to our H-bonding described in Scopino et al. (2020) paper, so GUU only was used to test 2x1ns, 1ns, and 500ps. The results indicated that the optimal protocol was likely a single piggyback step between 0 - 100ps. 10ps, 25ps, 50ps, 75ps, and 100ps were tested for GAU, GCU, GGU, and GUU, with the results summarized below.

Run Stability

The percentages of runs that ran to completion are indicated above. As long as the other analyses below indicate that our piggyback structures aren't significantly different from the no piggyback runs, we can decide on the length of the piggyback protocol from this information. Because we want to match the 80% - 100% completion of the other stages, this indicates that a single step 50ps piggyback is the best choice.

H-bond Depression Significance

Replication of the main results from the Scopino et al. (2020) paper is the most important criteria for deciding if the piggyback makes the CAR functionally different. The key finding is that the GAU and GGU codons have decreased H-bonding to the CAR (p < 0.01, t-test). The 50ps piggyback is in agreement with this, and also shows that the GUU codon has depressed H-bonding (p < 0.01). 50ps remains a valid choice of piggyback length.

RMSD

RMSD measures how far (in angstroms), on average, each atom in a structure has deviated relative to some reference. In this case, the reference is the energy minimized structure (emin11). While an RMSD <2A is acceptable, the smaller the value is the better. Having a piggyback step increases the RMSD, but the length of the piggyback over the range used does not have a large effect on RMSD. This is true for both backbone residues in the onion shell and backbone residues outside of the onion shell. 50ps remains a valid choice of piggyback length.

Energy Minimization

Finally, energy minimization was run to indirectly get a sense of how different each piggyback structure was to the energy minimized structure (emin11). This involved putting a 10kcal restraint on the structures at the end of piggybacking (or equilibration, as our control), to emin11. The overall energy levels were similar to the old protocol, with a larger deviation from the emin11 structure as the length of the piggyback increased. Similar to the RMSD, all of the piggyback protocols produce results within a reasonable range. 50ps, therefore, remains a valid choice of piggyback length and will be used going forward with our experiments with 5JUP.

Protocol

Piggybacking sample input

   50ps piggyback
    &cntrl
     imin     = 0,    	! no minimization
     ntx      = 5,    	! velocities inherited
     irest    = 1,    	! velocities inherited
     ntpr     = 5000, 	! print energy info every `ntpr` steps
     ntwr     = 50000,	! rewrite rst file every `ntwr` steps
     ntwx     = 5000, 	! write coord to trj every `ntwx` steps
     ntf      = 2,    	! bond interactions involving H omitted
     ntc      = 2,    	! SHAKE on, Hbonds constrained
     cut      = 8.0,  	! non-bond cutoff of 8A
     ntb      = 2,    	! 2 periodic boundaries for constant pressure
     nstlim   = 25000,	! number of MD steps to be performed (50ps)
     dt       = 0.002,	! time step in psec
     tempi    = 0.0,  	! initial temperature
     temp0    = 300,  	! ref temperature
     ntt      = 3,    	! Langevin dynamics 
     gamma_ln = 1.0,  	! collision freq 
     ntp      = 1,    	! constant pressure dynamics
     pres0    = 1.0,  	! reference pressure 1
     taup     = 5.0,  	! time constant for pressure
     nmropt   = 1,    	! restraint on
     ioutfm   = 1,    	! write binary trajectory 
     ntr      = 1,    	! restraint on
     restraint_wt = 20.0,
     restraintmask=':1,6-12,17-22,32-45,52-64,74,77-87,105-121,142-163,
        176-183,196,199-221,224-229,247-255,270-287,290-319,331-345,363-369,
        390-400,416-419,431-443,453-460,469-479,490-494',
     /
    &end
    &wt
     type='END',
    &end

Back to N1