- Posts: 3
binding energy
- Serrano
-
Topic Author
- Offline
- Fresh Boarder
Less
More
2 years 11 months ago #8941
by Serrano
binding energy was created by Serrano
In Section "2B. Add the ligand to the simulation box" of the
Small molecule binding tutorial
, it says "at the end of the file (just before the cell length), add the coordinates of the ligand from the file benzene.gro". Where is that file? I followed all the previous steps in this tutorial and it didn't generate that file with the coordinates from the ligand. Thx
Please Log in or Create an account to join the conversation.
- riccardo
-
- Offline
- Platinum Boarder
2 years 11 months ago #8942
by riccardo
Replied by riccardo on topic binding energy
Maybe it's not included. Anyway, you can generate it in several ways. From the benzene M3 model topology that you can find in
cgmartini.nl/images/martini_v300.zip
(the "small_molecule" file), you can see that it's a 3-bead model. Once you know that, for example, you can upload a benzene.pdb (atomistic) to
jbarnoud.github.io/cgbuilder/
and create a 3 bead mapping by clicking on the atoms and then downloading the generated GRO file. Or you could just draw 3 atoms at about 0.3 nm distance from each other with whatever other software (the bond length of benzene, see
cgmartini.nl/images/martini_v300.zip
, in the "small_molecule" file).
Please Log in or Create an account to join the conversation.
- Serrano
-
Topic Author
- Offline
- Fresh Boarder
Less
More
- Posts: 3
2 years 11 months ago - 2 years 11 months ago #8950
by Serrano
Replied by Serrano on topic binding energy
Thanks for your answer,
Riccardo
. My field of expertise is on Reinforcement Learning (not Molecular Dynamics), though. So, I don't exactly know how to create the 3-bead model. Once I upload the benzene in the
CG BuilderCG Builder
, should I add just one atom for each bead? Should those beads be adjacent? Anyway, I'll get help from a expert.
More importantly, and beyond this Docking tutorial, I'd like to know a couple of things to determine if I could use your Martini CG energy function for my research:
More importantly, and beyond this Docking tutorial, I'd like to know a couple of things to determine if I could use your Martini CG energy function for my research:
- I'd like to get the overall binding score from the Martini CG energy function for a given position between the ligand and the protein. Since I couldn't finished the aforementioned tutorial, I don't know if the output files provide such information. Could you confirm that?
- I'd like to use the Martini's binding score as a reward signal during a Reinforcement Learning training to perform docking. The protein would remain fixed while the ligand would move and rotate in a discrete molecular space. The computation of the Martini's binding score would be performed constantly during the training process. Would I need to convert the atomistic ligand structure into a CG model and to generate the Martini topology in each timestep considering that the ligand is constantly moving? Could the computation of the binding energy in each timestep be performed quickly (< 1 second)?
Last edit: 2 years 11 months ago by Serrano.
Please Log in or Create an account to join the conversation.
- riccardo
-
- Offline
- Platinum Boarder
2 years 10 months ago #8959
by riccardo
Replied by riccardo on topic binding energy
A benzene.gro:
Every bead in the Martini 3 benzene model maps to 2 carbons and 2 hydrogens of the atomistic structure.
Regarding your questions, they are not so trivial, at least to me. I'm not the ligand-protein binding expert here but let me try to address them:
1. in short, no, a Martini MD won't just spit out a binding score. You can get a lot of information, though, out of the MD simulation which could be used as a "binding score". The tutorial - which you can now hopefully go through - shows you how to get the k_on. Given that you sample exhaustively, you can also compute the binding free energy, as done in the paper: www.nature.com/articles/s41467-020-17437-5 (Figure 1 and Table 1). However given your 2nd point, I don't think these are the kind of quantities you want to compute, because they require info from long MD (they include the entropy).
2. No need to go back and forth between atomistic, or all-atom (AA) and Martini. You can do everything you want at the martini level, you'll just rely on the Martini force field instead of the AA force field. So things will be more approximate - but sampling will be faster - than AA. For the rest, basically anything you can compute at the AA level you can compute with Martini. What you are looking for is like some estimation of an enthalpic binding energy. E.g., you could compute the interaction energy between the ligand and the rest of the environment (protein + solvent) and this would be very quick to do. What you have to think carefully about though is whether that is what you need.
Hope this helps.
One benzene molecule
3
1BENZ R1 1 3.092 0.660 2.416 -0.0281 -0.1570 -0.0973
1BENZ R2 2 2.882 0.500 2.361 -0.1550 0.0200 -0.1344
1BENZ R3 3 2.986 0.658 2.168 -0.0360 -0.0103 -0.0960
4.04705 4.04705 4.04705Every bead in the Martini 3 benzene model maps to 2 carbons and 2 hydrogens of the atomistic structure.
Regarding your questions, they are not so trivial, at least to me. I'm not the ligand-protein binding expert here but let me try to address them:
1. in short, no, a Martini MD won't just spit out a binding score. You can get a lot of information, though, out of the MD simulation which could be used as a "binding score". The tutorial - which you can now hopefully go through - shows you how to get the k_on. Given that you sample exhaustively, you can also compute the binding free energy, as done in the paper: www.nature.com/articles/s41467-020-17437-5 (Figure 1 and Table 1). However given your 2nd point, I don't think these are the kind of quantities you want to compute, because they require info from long MD (they include the entropy).
2. No need to go back and forth between atomistic, or all-atom (AA) and Martini. You can do everything you want at the martini level, you'll just rely on the Martini force field instead of the AA force field. So things will be more approximate - but sampling will be faster - than AA. For the rest, basically anything you can compute at the AA level you can compute with Martini. What you are looking for is like some estimation of an enthalpic binding energy. E.g., you could compute the interaction energy between the ligand and the rest of the environment (protein + solvent) and this would be very quick to do. What you have to think carefully about though is whether that is what you need.
Hope this helps.
Please Log in or Create an account to join the conversation.
- d301108007@tmu.edu.tw
-
- Offline
- Fresh Boarder
Less
More
- Posts: 3
2 years 8 months ago #9097
by d301108007@tmu.edu.tw
Replied by d301108007@tmu.edu.tw on topic binding energy
Hi,
1. Is it possible to run ligand-protein simulation using Martini2, instead of Martini3?
2. How do we get the values to calculate the binding energy from ligand-protein simulation? Thanks in advance.
1. Is it possible to run ligand-protein simulation using Martini2, instead of Martini3?
2. How do we get the values to calculate the binding energy from ligand-protein simulation? Thanks in advance.
Please Log in or Create an account to join the conversation.
- riccardo
-
- Offline
- Platinum Boarder
2 years 7 months ago #9114
by riccardo
Replied by riccardo on topic binding energy
1. Yeah, but I would strongly recommend Martini 3. See discussion session of this paper:
www.nature.com/articles/s41467-020-17437-5.pdf
(page 8-9) for a discussion about improvements in Martini 3 over 2 in particular in this regard.
2. There's some info above, plus this new tutorial may give you some more ideas: cgmartini.nl/index.php/2021-martini-onli...all-molecule-binding
2. There's some info above, plus this new tutorial may give you some more ideas: cgmartini.nl/index.php/2021-martini-onli...all-molecule-binding
Please Log in or Create an account to join the conversation.
Time to create page: 0.096 seconds