Closed GlebNovikov closed 7 years ago
Joao, maybe this deserves a tutorial? I have something for visualizing 2d reaction coordinates
On 4 January 2017 at 12:15, GlebNovikov notifications@github.com wrote:
Dear HTMD users!
I am interesting to test various dimensionality reduction methods, as the alternative to PCA, for my set of MD trajectories of GPCR systems. I would like to cluster all trajectories onto the 2D plane of shared collective variables (meaning that it was calculated from all ensemble of md snapshots) to i) make visualisation of the collective deformations responsible for some biological activity of the GPCR observed in MD; ii) to monitor how the dynamical equilibrium is shifted along those CV in each of the system separately (e.g depending on the presence of various ligands in each case), thus obtaining slice of free energy surface for my systems along CVs or just to make projections of each trajectories separately.
Methodologically, I have prepared my MD data in the manner how I do it for PCA: In brief, I have md three trajectories in dcd format, each consisted of same number of atoms and equal number of snapshots, meaning that share same duration, consisting of only receptors atoms in each case (without ligand, solvent membrane etc) + the pdb of GPCRs with the same number of atoms.
Now in TICA tutorial I don't understand clearly how to load all of those dcd trajectories, fit it to the reference pdb and properly select CVs for the clustering (it's better to make some visualization of it in VMD or another software).
I would be especially thankful for any help or some template script suitable for my task!
Gleb
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/Acellera/htmd/issues/215, or mute the thread https://github.com/notifications/unsubscribe-auth/AHaqOs6kJZHhXDR3tenatPYgIxsoi45tks5rO39BgaJpZM4LahXA .
<https://twitter.com/acellera>
https://www.youtube.com/user/acelleracom https://www.linkedin.com/company/2133167?trk=tyah&trkInfo=clickedVertical%3Acompany%2CclickedEntityId%3A2133167%2Cidx%3A2-1-2%2CtarId%3A1448018583204%2Ctas%3Aacellera https://www.acellera.com/md-simulation-blog-news/ http://is.gd/1eXkbS
Thanks so much for the responce, Gianni! Indeed such tutorial on the application of the TICA for clustering of md trajectories along two chosen collective coordinates and subsequent visualisation of the projections will be very useful! I did it many times using PCA implemented in prody for example - which is another python-based package- if it will be interesting I will send you link on the tutorial.
Gleb
sure, @giadefa, can you send me your example?
Hi folks, we're working on something similar here and will be releasing soon. So perhaps it's worth looking at it together
hi @gph82! that would be a good idea. I am going to be at the PyEMMA workshop in February, maybe we can even discuss it then?
that'd be great, we'll all be here
On 09.01.2017 12:01, João M. Damas wrote:
hi @gph82 https://github.com/gph82! that would be a good idea. I am going to be at the PyEMMA workshop in February, maybe we can even discuss it then?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/Acellera/htmd/issues/215#issuecomment-271257906, or mute the thread https://github.com/notifications/unsubscribe-auth/AHK3NO5re6WKKV1RGSpDlxLg0_38MKpYks5rQhN4gaJpZM4LahXA.
isn't this very simple?
On 9 January 2017 at 12:01, João M. Damas notifications@github.com wrote:
hi @gph82 https://github.com/gph82! that would be a good idea. I am going to be at the PyEMMA workshop in February, maybe we can even discuss it then?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/Acellera/htmd/issues/215#issuecomment-271257906, or mute the thread https://github.com/notifications/unsubscribe-auth/AHaqOnjTMjTrJKSIkT48Q2vLViY3nPGzks5rQhN4gaJpZM4LahXA .
<https://twitter.com/acellera>
https://www.youtube.com/user/acelleracom https://www.linkedin.com/company/2133167?trk=tyah&trkInfo=clickedVertical%3Acompany%2CclickedEntityId%3A2133167%2Cidx%3A2-1-2%2CtarId%3A1448018583204%2Ctas%3Aacellera https://www.acellera.com/md-simulation-blog-news/ http://is.gd/1eXkbS
depends on which things we want to plot, i guess
free energies
On 9 January 2017 at 12:21, João M. Damas notifications@github.com wrote:
depends on which things we want to plot, i guess
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/Acellera/htmd/issues/215#issuecomment-271261577, or mute the thread https://github.com/notifications/unsubscribe-auth/AHaqOixNDsPg_9AaEI8nC4e57nKvEXWTks5rQhhQgaJpZM4LahXA .
<https://twitter.com/acellera>
https://www.youtube.com/user/acelleracom https://www.linkedin.com/company/2133167?trk=tyah&trkInfo=clickedVertical%3Acompany%2CclickedEntityId%3A2133167%2Cidx%3A2-1-2%2CtarId%3A1448018583204%2Ctas%3Aacellera https://www.acellera.com/md-simulation-blog-news/ http://is.gd/1eXkbS
with the point in the tIC space and some kerneling, it should be easy to create a conditional free energy surface, yes. But I would guess that would be a very short tutorial, and it would be nice to complement it a bit more.
Hi @giadefa, yes it's easy. But depending on the level of automation you want there are some practical api functions we're working on.
Hello and many thanks for the attention to my topic again!
Yes, the general idea is to compute free energy as Boltzmann probability densities along projections of md trajectories onto the 2D plane of collective coordinates computed by TICO - i.e via piloting a histogram - what possible to do by means of PCA in GROMACs via following combinations of GMX utilities g_covar g_anaeig g_shame
tutorial: http://www3.mpibpc.mpg.de/groups/de_groot/compbio/p4/index.html
For the biologists such method helps to "track" in which conformations protein spent more time during md simulations i.e compared typical system with different bound ligands or mutations - see how it redistribute collective equilibrium. Assuming that those collective deformations are functionally important being low-energy pathways to deform the structure for achievement of it function - the TICO will useful both as a tool for for the visualization as well as for the analysis of protein activity.
Gleb
i'll pass the code to Joao, but it's very simple.
On 10 January 2017 at 15:28, GlebNovikov notifications@github.com wrote:
Hello and many thanks for the attention to my topic again!
Yes, the general idea is to compute free energy as Boltzmann probability densities along projections of md trajectories onto the 2D plane of collective coordinates computed by TICO - i.e via piloting a histogram - what possible to do by means of PCA in GROMACs via following combinations of GMX utilities g_covar g_anaeig g_shame tutorial: http://www3.mpibpc.mpg.de/groups/de_groot/compbio/p4/index.html
For the biologists such method helps to "track" in which conformations protein spent more time during md simulations i.e compared typical system with different bound ligands or mutations - see how it redistribute collective equilibrium. Assuming that those collective deformations are functionally important being low-energy pathways to deform the structure for achievement of it function - the TICO will useful both as a tool for for the visualization as well as for the analysis of protein activity.
Gleb
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/Acellera/htmd/issues/215#issuecomment-271589003, or mute the thread https://github.com/notifications/unsubscribe-auth/AHaqOgD7X3eroB1TYaYFbW4MMjSwrE5Yks5rQ5WPgaJpZM4LahXA .
<https://twitter.com/acellera>
https://www.youtube.com/user/acelleracom https://www.linkedin.com/company/2133167?trk=tyah&trkInfo=clickedVertical%3Acompany%2CclickedEntityId%3A2133167%2Cidx%3A2-1-2%2CtarId%3A1448018583204%2Ctas%3Aacellera https://www.acellera.com/md-simulation-blog-news/ http://is.gd/1eXkbS
@GlebNovikov I added functionality for this in MetricData.plotCounts as well as Model.plotFES
Dear HTMD users!
I am interesting to test various dimensionality reduction methods, as the alternative to PCA, for my set of MD trajectories of GPCR systems. I would like to cluster all trajectories onto the 2D plane of shared collective variables (meaning that it was calculated from all ensemble of md snapshots) to i) make visualisation of the collective deformations responsible for some biological activity of the GPCR observed in MD; ii) to monitor how the dynamical equilibrium is shifted along those CV in each of the system separately (e.g depending on the presence of various ligands in each case), thus obtaining slice of free energy surface for my systems along CVs or just to make projections of each trajectories separately.
Methodologically, I have prepared my MD data in the manner how I do it for PCA: In brief, I have md three trajectories in dcd format, each consisted of same number of atoms and equal number of snapshots, meaning that share same duration, consisting of only receptors atoms in each case (without ligand, solvent membrane etc) + the pdb of GPCRs with the same number of atoms.
Now in TICA tutorial I don't understand clearly how to load all of those dcd trajectories, fit it to the reference pdb and properly select CVs for the clustering (it's better to make some visualization of it in VMD or another software).
I would be especially thankful for any help or some template script suitable for my task!
Gleb