Coloring based on confidence

[emehinovic72]

Is there a way to set the color scheme to reflect on PLDDT confidence for those proteins structures outputted by Alphafold?

[appassionate]

Hi, I have similar problems on it. I tried to applied a array with (n_frames, n_atoms) shape to reflect a "condition" per frame.
I tried to set a "color_scheme" to a Widget class, however "_ColormakerRegistry" seems not work in a initialized widget object. Maybe, I want to know more about "_remote_call" to control those embedded JS...
Also, if there is any a method to let "widget" get an extra data for this kind representations?
Any ideas will be inspired me for i am just a greenhand. : )

[emehinovic72]

view.add_cartoon('protein',color_scheme='bfactor') is close, but the color scale needs to be adjusted 

[hainm]

view.add_cartoon('protein',color_scheme='bfactor') is close, but the color scale needs to be adjusted 

@emehinovic72 Maybe try customizing the color scheme yourself? #839

[hainm]

Hi, I have similar problems on it. I tried to applied a array with (n_frames, n_atoms) shape to reflect a "condition" per frame. I tried to set a "color_scheme" to a Widget class, however "_ColormakerRegistry" seems not work in a initialized widget object. Maybe, I want to know more about "_remote_call" to control those embedded JS... Also, if there is any a method to let "widget" get an extra data for this kind representations? Any ideas will be inspired me for i am just a greenhand. : )

hi @appassionate I don't clearly understand your issue, but if you want to update the color based on frame changed, you can play with this code:

from time import sleep

def on_change(change):
    frame = change.new
    if frame % 2:
        color = 'red'
    else:
        color = 'blue'
    view.update_cartoon(color=color)
    sleep(0.1) # wait for the color update
view.observe(on_change, names=['frame'])

[emehinovic72]

view.add_cartoon('protein',color_scheme='bfactor') is close, but the color scale needs to be adjusted 

@emehinovic72 Maybe try customizing the color scheme yourself? #839

Would you be able to provide an example(s) for color customization by bfactor?

[appassionate]

Hi, I have similar problems on it. I tried to applied a array with (n_frames, n_atoms) shape to reflect a "condition" per frame. I tried to set a "color_scheme" to a Widget class, however "_ColormakerRegistry" seems not work in a initialized widget object. Maybe, I want to know more about "_remote_call" to control those embedded JS... Also, if there is any a method to let "widget" get an extra data for this kind representations? Any ideas will be inspired me for i am just a greenhand. : )

hi @appassionate I don't clearly understand your issue, but if you want to update the color based on frame changed, you can play with this code:

from time import sleep

def on_change(change):
    frame = change.new
    if frame % 2:
        color = 'red'
    else:
        color = 'blue'
    view.update_cartoon(color=color)
    sleep(0.1) # wait for the color update
view.observe(on_change, names=['frame'])

Thanks! But "update_cartoon" seems not work in my work.. maybe I even have not used "add_cartoon" in my viewer.
Actually, I followed class "NGLDisplay" in ase. It represented atoms by using "add_spacefill", however "update_spacefill"
seems changed all color in spacefill representation instead of"changing a selected color of one of atoms " during my attempt... :(
In my naive try: I want to create "spacefill" for every atoms, it will cost a lot times...

n_frames = 100
atom_num = len(u.atoms)
value_array = np.random.randint(0,3,(n_frames,atom_num))
color_list = ['red', 'blue', 'yellow']

for i in range(n_frames):
    viewer.view.remove_spacefill()
    for j in range(atom_num):
        _color = color_list[value_array[i][j]]
        viewer.view.add_spacefill(selection=f'@{j}',radiusScale=2,color=f'{_color}')
    sleep(1)

[hainm]

Thanks! But "update_cartoon" seems not work in my work.. maybe I even have not used "add_cartoon" in my viewer.

hi @appassionate That's a copy and paste code for you to adapt it. You need to write the on_change function by yourself.

In my naive try: I want to create "spacefill" for every atoms, it will cost a lot times...

where does the bottleneck come from?

[hainm]

view.add_cartoon('protein',color_scheme='bfactor') is close, but the color scale needs to be adjusted 

@emehinovic72 Maybe try customizing the color scheme yourself? #839

Would you be able to provide an example(s) for color customization by bfactor?

I am not sure "how". But here is the JS code: https://github.com/nglviewer/ngl/blob/master/src/color/bfactor-colormaker.ts

In this PR, I showed an example of how to write your own JS code to customize your color. (I admit I am more familiar with Python): #839 (See Function section)

[appassionate]

Thanks! But "update_cartoon" seems not work in my work.. maybe I even have not used "add_cartoon" in my viewer.

hi @appassionate That's a copy and paste code for you to adapt it. You need to write the on_change function by yourself.

In my naive try: I want to create "spacefill" for every atoms, it will cost a lot times...

where does the bottleneck come from?

Many thanks for your reply.. I believe I have learned about "add_cartoon" will be a "protein representation" which is not related to my "atom representation".
I think "add_spacefill"(a kind of add_representation? ) for every atoms will lead to a big burden for "rendering engine"(I guess it exists in JS codes somewhere). One frame for atoms will be trouble let alone the rendering of each frame... It is a bad try....
So, I continued to try "update_spacefill" by color, but I could not select atoms exactly, but to have selcected all atoms in a spacefill... I guess it is a "generated spacefill object" that cant be changed anymore by "remote_call" method, or is there any keyargs for changing a generated spacefill by atom index?

I believe it will be easier to set a ColorScheme to represent a kind of property such as "bfactors". However, the property value in my task is not in trajectory files but a numpy array... It seems that there is less api between JS and python.... it will be a complexed task compared with my imaged "update_spacefill" way.
maybe there are some things i should do....

1. find some way like nglview.adaptor.structure using to load my value, like "def get_coordinates" and find or develop an api to communicate with JS code.
2. just like atoms.serial in JS code, we may have "atom.so_called_value" for every frame as a streaming reading.
3. a registered colorscheme to let "render engine" know to give balls or other topology object colors.

Anyway, thanks for your help. :p

[hainm]

Cheers.