Dear ChimeraX developers, I have been wondering about the pros/cons between the 2 methods to smoothen a trajectory. If I provide a pdb and an nc trajectory file, I can use the smooth_md python script. If the trajectory originally has 100 frames , there will still be 100 frames after using the script. If I exported my trajectory into separate pdb models and morph between each of them, with the default of 50 frames created between each trajectory frame, there will be 5,001 frames after using the command. While importing many pdb's and morphing them takes longer than importing a trajectory and smoothing it, are there any benefits to using either one over the other? I remember seeing some post-production artifacts in smooth_md python script method like the hydrogens of a lysine being in an odd geometry briefly throughout the smoothed trajectory and am unsure if morph command would do the same. Regards, Alex
Hi Alex, Maybe others have more specific answers, but my general sense is that the choice depends on artistic license + heatlthy awareness that any smoothing will alter the coordinates from what was produced by the energy function in your simulation. The ChimeraX smooth_md script (for others, see https://www.rbvi.ucsf.edu/pipermail/chimerax-users/2019-July/000590.html and the actual script is linked near the bottom of https://www.cgl.ucsf.edu/trac/chimera/wiki/Scripts ) averages coordinates in a frame based on the surrounding frames in a specified window width. If that is simply doing Cartesian averaging (sorry I'm not sure about that and can't read code) it is possible that would give some distortions if there are large changes across your specified window. The results you get from morphing of course depend on the options you chose. <https://rbvi.ucsf.edu/chimerax/docs/user/commands/morph.html> More frames => more opportunity to deviate from what the energy function gave you, but at least the default (cartesian false) tries to interpolate the internal coordinates where possible to avoid ridiculous local conformations. It also depends on your trajectory time step and at what temperature the simulation was run (how large the changes you observe between two adjacent trajectory frames might be). My intuition is that 50 frames would be rather excessive for smoothing between each pair of frames of trajectory, and you might want to try something much smaller (3? 5?), but you'd have to see how it looks. The defaults are what we felt reasonable for the most common use case of comparing conformational changes of moderate size between two different experimentally determined structures, not for smoothing trajectories. I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Resource for Biocomputing, Visualization, and Informatics Department of Pharmaceutical Chemistry University of California, San Francisco
On May 19, 2025, at 9:26 AM, Alexander Lee via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Dear ChimeraX developers,
I have been wondering about the pros/cons between the 2 methods to smoothen a trajectory.
If I provide a pdb and an nc trajectory file, I can use the smooth_md python script. If the trajectory originally has 100 frames , there will still be 100 frames after using the script.
If I exported my trajectory into separate pdb models and morph between each of them, with the default of 50 frames created between each trajectory frame, there will be 5,001 frames after using the command.
While importing many pdb's and morphing them takes longer than importing a trajectory and smoothing it, are there any benefits to using either one over the other? I remember seeing some post-production artifacts in smooth_md python script method like the hydrogens of a lysine being in an odd geometry briefly throughout the smoothed trajectory and am unsure if morph command would do the same.
Regards, Alex
Hi Elaine, Thank you for your insight into the 2 techniques. I will keep in mind to drastically lower the frame # option when using morph in the future for trajectories. Is there code that shows the implementation of corkscrew interpolation in the morph command? I want to see if a derivative of the smooth_MD script is possible with interpolation instead of cartesian averaging. Another question I have is about post-smoothing analysis on simulations. With the distortions created through either smoothing methods, might they impact analyses like finding H-bonds? Regards, Alex ________________________________ From: Elaine Meng <meng@cgl.ucsf.edu> Sent: Monday, May 19, 2025 2:35 PM To: Alexander Lee <Alex.H.Lee@outlook.com> Cc: Alexandra Zahradnikova via ChimeraX-users <chimerax-users@cgl.ucsf.edu> Subject: Re: [chimerax-users] Smooth MD script vs morph movie Hi Alex, Maybe others have more specific answers, but my general sense is that the choice depends on artistic license + heatlthy awareness that any smoothing will alter the coordinates from what was produced by the energy function in your simulation. The ChimeraX smooth_md script (for others, see https://www.rbvi.ucsf.edu/pipermail/chimerax-users/2019-July/000590.html and the actual script is linked near the bottom of https://www.cgl.ucsf.edu/trac/chimera/wiki/Scripts ) averages coordinates in a frame based on the surrounding frames in a specified window width. If that is simply doing Cartesian averaging (sorry I'm not sure about that and can't read code) it is possible that would give some distortions if there are large changes across your specified window. The results you get from morphing of course depend on the options you chose. <https://rbvi.ucsf.edu/chimerax/docs/user/commands/morph.html> More frames => more opportunity to deviate from what the energy function gave you, but at least the default (cartesian false) tries to interpolate the internal coordinates where possible to avoid ridiculous local conformations. It also depends on your trajectory time step and at what temperature the simulation was run (how large the changes you observe between two adjacent trajectory frames might be). My intuition is that 50 frames would be rather excessive for smoothing between each pair of frames of trajectory, and you might want to try something much smaller (3? 5?), but you'd have to see how it looks. The defaults are what we felt reasonable for the most common use case of comparing conformational changes of moderate size between two different experimentally determined structures, not for smoothing trajectories. I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Resource for Biocomputing, Visualization, and Informatics Department of Pharmaceutical Chemistry University of California, San Francisco
On May 19, 2025, at 9:26 AM, Alexander Lee via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Dear ChimeraX developers,
I have been wondering about the pros/cons between the 2 methods to smoothen a trajectory.
If I provide a pdb and an nc trajectory file, I can use the smooth_md python script. If the trajectory originally has 100 frames , there will still be 100 frames after using the script.
If I exported my trajectory into separate pdb models and morph between each of them, with the default of 50 frames created between each trajectory frame, there will be 5,001 frames after using the command.
While importing many pdb's and morphing them takes longer than importing a trajectory and smoothing it, are there any benefits to using either one over the other? I remember seeing some post-production artifacts in smooth_md python script method like the hydrogens of a lysine being in an odd geometry briefly throughout the smoothed trajectory and am unsure if morph command would do the same.
Regards, Alex
Hi Alex, Any change in coordinates can affect analyses that use those coordinates (hbonds, contacts, surface areas ...) As for code, somebody else will have to answer as to where exactly to look. I'm the nonprogrammer of the group. Best, Elaine
On May 19, 2025, at 1:30 PM, Alexander Lee via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi Elaine,
Thank you for your insight into the 2 techniques. I will keep in mind to drastically lower the frame # option when using morph in the future for trajectories. Is there code that shows the implementation of corkscrew interpolation in the morph command? I want to see if a derivative of the smooth_MD script is possible with interpolation instead of cartesian averaging.
Another question I have is about post-smoothing analysis on simulations. With the distortions created through either smoothing methods, might they impact analyses like finding H-bonds?
Regards, AlexFrom: Elaine Meng <meng@cgl.ucsf.edu> Sent: Monday, May 19, 2025 2:35 PM To: Alexander Lee <Alex.H.Lee@outlook.com> Cc: Alexandra Zahradnikova via ChimeraX-users <chimerax-users@cgl.ucsf.edu> Subject: Re: [chimerax-users] Smooth MD script vs morph movie Hi Alex, Maybe others have more specific answers, but my general sense is that the choice depends on artistic license + heatlthy awareness that any smoothing will alter the coordinates from what was produced by the energy function in your simulation.
The ChimeraX smooth_md script (for others, see https://www.rbvi.ucsf.edu/pipermail/chimerax-users/2019-July/000590.html and the actual script is linked near the bottom ofhttps://www.cgl.ucsf.edu/trac/chimera/wiki/Scripts ) averages coordinates in a frame based on the surrounding frames in a specified window width. If that is simply doing Cartesian averaging (sorry I'm not sure about that and can't read code) it is possible that would give some distortions if there are large changes across your specified window.
The results you get from morphing of course depend on the options you chose. <https://rbvi.ucsf.edu/chimerax/docs/user/commands/morph.html> More frames => more opportunity to deviate from what the energy function gave you, but at least the default (cartesian false) tries to interpolate the internal coordinates where possible to avoid ridiculous local conformations. It also depends on your trajectory time step and at what temperature the simulation was run (how large the changes you observe between two adjacent trajectory frames might be).
My intuition is that 50 frames would be rather excessive for smoothing between each pair of frames of trajectory, and you might want to try something much smaller (3? 5?), but you'd have to see how it looks. The defaults are what we felt reasonable for the most common use case of comparing conformational changes of moderate size between two different experimentally determined structures, not for smoothing trajectories.
I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Resource for Biocomputing, Visualization, and Informatics Department of Pharmaceutical Chemistry University of California, San Francisco
On May 19, 2025, at 9:26 AM, Alexander Lee via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Dear ChimeraX developers,
I have been wondering about the pros/cons between the 2 methods to smoothen a trajectory.
If I provide a pdb and an nc trajectory file, I can use the smooth_md python script. If the trajectory originally has 100 frames , there will still be 100 frames after using the script.
If I exported my trajectory into separate pdb models and morph between each of them, with the default of 50 frames created between each trajectory frame, there will be 5,001 frames after using the command.
While importing many pdb's and morphing them takes longer than importing a trajectory and smoothing it, are there any benefits to using either one over the other? I remember seeing some post-production artifacts in smooth_md python script method like the hydrogens of a lysine being in an odd geometry briefly throughout the smoothed trajectory and am unsure if morph command would do the same.
Regards, Alex
_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
Hi Alex, The interpolation code is in the morph <https://github.com/RBVI/ChimeraX/tree/develop/src/bundles/morph/src> module, particularly the interpolate.py and interp_residue.py files. It looks pretty complicated. The programmer who wrote it (Tom Goddard) is on vacation until Thursday. --Eric Eric Pettersen UCSF Computer Graphics Lab
On May 19, 2025, at 1:36 PM, Elaine Meng via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi Alex, Any change in coordinates can affect analyses that use those coordinates (hbonds, contacts, surface areas ...)
As for code, somebody else will have to answer as to where exactly to look. I'm the nonprogrammer of the group. Best, Elaine
On May 19, 2025, at 1:30 PM, Alexander Lee via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi Elaine,
Thank you for your insight into the 2 techniques. I will keep in mind to drastically lower the frame # option when using morph in the future for trajectories. Is there code that shows the implementation of corkscrew interpolation in the morph command? I want to see if a derivative of the smooth_MD script is possible with interpolation instead of cartesian averaging.
Another question I have is about post-smoothing analysis on simulations. With the distortions created through either smoothing methods, might they impact analyses like finding H-bonds?
Regards, AlexFrom: Elaine Meng <meng@cgl.ucsf.edu> Sent: Monday, May 19, 2025 2:35 PM To: Alexander Lee <Alex.H.Lee@outlook.com> Cc: Alexandra Zahradnikova via ChimeraX-users <chimerax-users@cgl.ucsf.edu> Subject: Re: [chimerax-users] Smooth MD script vs morph movie Hi Alex, Maybe others have more specific answers, but my general sense is that the choice depends on artistic license + heatlthy awareness that any smoothing will alter the coordinates from what was produced by the energy function in your simulation.
The ChimeraX smooth_md script (for others, see https://www.rbvi.ucsf.edu/pipermail/chimerax-users/2019-July/000590.html and the actual script is linked near the bottom ofhttps://www.cgl.ucsf.edu/trac/chimera/wiki/Scripts ) averages coordinates in a frame based on the surrounding frames in a specified window width. If that is simply doing Cartesian averaging (sorry I'm not sure about that and can't read code) it is possible that would give some distortions if there are large changes across your specified window.
The results you get from morphing of course depend on the options you chose. <https://rbvi.ucsf.edu/chimerax/docs/user/commands/morph.html> More frames => more opportunity to deviate from what the energy function gave you, but at least the default (cartesian false) tries to interpolate the internal coordinates where possible to avoid ridiculous local conformations. It also depends on your trajectory time step and at what temperature the simulation was run (how large the changes you observe between two adjacent trajectory frames might be).
My intuition is that 50 frames would be rather excessive for smoothing between each pair of frames of trajectory, and you might want to try something much smaller (3? 5?), but you'd have to see how it looks. The defaults are what we felt reasonable for the most common use case of comparing conformational changes of moderate size between two different experimentally determined structures, not for smoothing trajectories.
I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Resource for Biocomputing, Visualization, and Informatics Department of Pharmaceutical Chemistry University of California, San Francisco
On May 19, 2025, at 9:26 AM, Alexander Lee via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Dear ChimeraX developers,
I have been wondering about the pros/cons between the 2 methods to smoothen a trajectory.
If I provide a pdb and an nc trajectory file, I can use the smooth_md python script. If the trajectory originally has 100 frames , there will still be 100 frames after using the script.
If I exported my trajectory into separate pdb models and morph between each of them, with the default of 50 frames created between each trajectory frame, there will be 5,001 frames after using the command.
While importing many pdb's and morphing them takes longer than importing a trajectory and smoothing it, are there any benefits to using either one over the other? I remember seeing some post-production artifacts in smooth_md python script method like the hydrogens of a lysine being in an odd geometry briefly throughout the smoothed trajectory and am unsure if morph command would do the same.
Regards, Alex
_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
Hi Alex, The corkscrew interpolated coordinates are not differentiable at the timepoints of an MD trajectory if you morph between each consecutive times. You don't need to look at code to conclude that if you understand the description in the documentation of the interpolation https://www.cgl.ucsf.edu/chimerax/docs/user/commands/morph.html#morph-parame... The interpolation only involves the coordinates of the two consecutive time points that your morph between. So it of course cannot be differentiable between segments (t0,t1) and (t1,t2). To have smoothness at t1 would require the segment (t0,t1) to know and use the coordinates at t2 and likewise the (t1,t2) segment would need to know and use the coordinates at t0. In other words, smoothness at t1 needs to consider coordinates both at t0 and t2. The morph algorithm doesn't interpolate 3 times, it only handles two consecutive times so it cannot be differentiable at t1 (except by accident when the t0,t1 segment extrapolation coincidentally exactly arrives at the t2 coordinates). Tom
On May 19, 2025, at 1:49 PM, Eric Pettersen via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi Alex, The interpolation code is in the morph <https://github.com/RBVI/ChimeraX/tree/develop/src/bundles/morph/src> module, particularly the interpolate.py and interp_residue.py files. It looks pretty complicated. The programmer who wrote it (Tom Goddard) is on vacation until Thursday.
--Eric
Eric Pettersen UCSF Computer Graphics Lab
On May 19, 2025, at 1:36 PM, Elaine Meng via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi Alex, Any change in coordinates can affect analyses that use those coordinates (hbonds, contacts, surface areas ...)
As for code, somebody else will have to answer as to where exactly to look. I'm the nonprogrammer of the group. Best, Elaine
On May 19, 2025, at 1:30 PM, Alexander Lee via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi Elaine,
Thank you for your insight into the 2 techniques. I will keep in mind to drastically lower the frame # option when using morph in the future for trajectories. Is there code that shows the implementation of corkscrew interpolation in the morph command? I want to see if a derivative of the smooth_MD script is possible with interpolation instead of cartesian averaging.
Another question I have is about post-smoothing analysis on simulations. With the distortions created through either smoothing methods, might they impact analyses like finding H-bonds?
Regards, AlexFrom: Elaine Meng <meng@cgl.ucsf.edu> Sent: Monday, May 19, 2025 2:35 PM To: Alexander Lee <Alex.H.Lee@outlook.com> Cc: Alexandra Zahradnikova via ChimeraX-users <chimerax-users@cgl.ucsf.edu> Subject: Re: [chimerax-users] Smooth MD script vs morph movie Hi Alex, Maybe others have more specific answers, but my general sense is that the choice depends on artistic license + heatlthy awareness that any smoothing will alter the coordinates from what was produced by the energy function in your simulation.
The ChimeraX smooth_md script (for others, see https://www.rbvi.ucsf.edu/pipermail/chimerax-users/2019-July/000590.html and the actual script is linked near the bottom ofhttps://www.cgl.ucsf.edu/trac/chimera/wiki/Scripts ) averages coordinates in a frame based on the surrounding frames in a specified window width. If that is simply doing Cartesian averaging (sorry I'm not sure about that and can't read code) it is possible that would give some distortions if there are large changes across your specified window.
The results you get from morphing of course depend on the options you chose. <https://rbvi.ucsf.edu/chimerax/docs/user/commands/morph.html> More frames => more opportunity to deviate from what the energy function gave you, but at least the default (cartesian false) tries to interpolate the internal coordinates where possible to avoid ridiculous local conformations. It also depends on your trajectory time step and at what temperature the simulation was run (how large the changes you observe between two adjacent trajectory frames might be).
My intuition is that 50 frames would be rather excessive for smoothing between each pair of frames of trajectory, and you might want to try something much smaller (3? 5?), but you'd have to see how it looks. The defaults are what we felt reasonable for the most common use case of comparing conformational changes of moderate size between two different experimentally determined structures, not for smoothing trajectories.
I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Resource for Biocomputing, Visualization, and Informatics Department of Pharmaceutical Chemistry University of California, San Francisco
On May 19, 2025, at 9:26 AM, Alexander Lee via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Dear ChimeraX developers,
I have been wondering about the pros/cons between the 2 methods to smoothen a trajectory.
If I provide a pdb and an nc trajectory file, I can use the smooth_md python script. If the trajectory originally has 100 frames , there will still be 100 frames after using the script.
If I exported my trajectory into separate pdb models and morph between each of them, with the default of 50 frames created between each trajectory frame, there will be 5,001 frames after using the command.
While importing many pdb's and morphing them takes longer than importing a trajectory and smoothing it, are there any benefits to using either one over the other? I remember seeing some post-production artifacts in smooth_md python script method like the hydrogens of a lysine being in an odd geometry briefly throughout the smoothed trajectory and am unsure if morph command would do the same.
Regards, Alex
_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
participants (4)
-
Alexander Lee -
Elaine Meng -
Eric Pettersen -
Tom Goddard