Re: [GaMD-discuss] LiGaMD3 parameters and reweightening
Brought to you by:
yinglongmiao
Menu
▾
▴
Re: [GaMD-discuss] LiGaMD3 parameters and reweightening
|
From: Marc C. J. <Mar...@ua...> - 2025-06-05 08:32:01
|
Hi Jinan, Thank you for your insights, I will continue testing with different parameters and doing longer simulations with LiGaMD3, this time starting with a completely unbound ligand. With igamd=11 though, then it cannot support semi-isotropic coupling of pressure and with a membrane system, for long simulations it’s important. Thanks again for your comments, Marc Ciruela From: 王进安 <jin...@ex...> Date: Wednesday, 4 June 2025 at 07:14 To: Miao, Yinglong <Yin...@me...> Cc: Marc Ciruela Jardí <Mar...@ua...>, gamd-discuss <gam...@li...>, Joshi, Keya <ke...@em...>, Adediwura, Victor Ayo <vad...@em...>, Xiao, Fu <Fu...@me...> Subject: Re: Re: [GaMD-discuss] LiGaMD3 parameters and reweightening No soleu rebre correu electrònic de jin...@ex.... descobriu per què aquest fet és important<https://aka.ms/LearnAboutSenderIdentification> Hi Marc, I'll try my best. I have been trying to run GaMD on a membrane protein system with a small molecule and found several issues I would appreciate your help with. First, since I am trying to characterize the binding of the small molecule to the protein, should I start with the ligand bound to the protein or unbound? I don’t have a completely defined binding mode still, so it would be an approximation. The ligand binds in the interface of the protein with the membrane lipids, so in case I put it unbound, should be already in the membrane probably… For the initial conformation, I think it is fine to put it unbound. Another doubt I have is what kind of biasing to apply. LiGaMD3 seems to fit the best, but because the membrane is also somehow involved in the binding, should I use another kind of GaMD? Maybe the dual boost GaMD where the boost is applied to all the system? For identifying the ligand binding pose, LiGaMD3 should be the best one. But you can also try the mode of igamd=11 and 15. I tried applying LiGaMD3 with the threshold in the upper bound (iEP=2, iED=2 but iEB=1) as indicated in the LiGaMD3 publication and with a series of sigma0P and sigma0D values, from 4 to 8 with sigma0B fixed at 6. To decide what values to use, I looked at the trajectories and observed that with sigma0P=sigma0D=4 I already found an unbinding event. Looking at the log, k0P already stabilized at 1 and k0D and k0B were near 1. The values of the last unboosted potential energy and unboosted dihedral energy where 30.238509140909 and -226779.050457056845. The dihedral energy is very small, right? Why may this be happening? I continued with a production run but never found a rebinding event in 1us of simulation. As mentioned in the paper, in order to capture both binding and dissociation, we should choose the parameter of sigma0P and sigma0D to the value of which the largest sigma0P to keep the ligand in the bound state after the equilibration and at the same time, keeping the iEP and iED in the mode of 2 in the final of the equilibration. The values is the first and second boost potential, so 30.23 and -226779 seems good to me. Despite this, I decided to try and analyse the simulation to calculate the PMF. The dV distribution was centered around 30-25 kcal, which seems quite high, but I suppose it’s because of iE=2. The distribution seemed normal, but a bit skewed. I tried doing a reweightening using several reaction coordinates but the profiles where very sharp, I suppose because of the lack of re-binding processes which gives a bad sampling. The boost potential of 25-30 kcal/mol is quite good as the selective GaMD often generate higher boost. For the sharp PMF, yes, it is the bad sampling issue. With the very low energetic values of the dihedral energy and high dV, I decided to try iE=1 but then, with the same sigma values I did not obtain any unbinding during the equilibration and the values for the potentials remained similar during the equilibration. That's normal as the iE=1 put much lower boost in the system. Finally I tried with igamd=3 setting iEP=iED=1 and sigma0P=sigma0D=6. Here, I also didn’t see any unbinding but interestingly, the last unboosted potential energy and unboosted dihedral energy in the equilibration where -184058.981408115011 17220.120062542148, so completely different. The energy here is the total and dihedral potential. So it is different in different mode. Previously I had run a GaMD with a similar system with a peptide instead of a small molecule, using Pep-GaMD and starting from a quasi-bound state I was able to find a stable binding mode for the peptide. So I would much appreaicate your insight on what may be happening to this system and what recommendations would you have. For this one, you might test the same Pep-GaMD. Best, Jinan -----原始邮件----- 发件人:"Miao, Yinglong" <Yin...@me...> 发送时间:2025-06-04 04:57:32 (星期三) 收件人: "Marc Ciruela Jardí" <Mar...@ua...> 抄送: gamd-discuss <gam...@li...>, "Jinan Wang" <jin...@ex...>, "Joshi, Keya" <ke...@em...>, "Adediwura, Victor Ayo" <vad...@em...>, "Xiao, Fu" <Fu...@me...> 主题: Re: [GaMD-discuss] LiGaMD3 parameters and reweightening Dear Marc, Thanks for your great questions. I’m CC’ing a number of LiGaMD3 experts and hope they will be able to share their insights for the discussion. Best regards, Yinglong Yinglong Miao, Ph.D. Associate Professor Department of Pharmacology and Computational Medicine Program University of North Carolina - Chapel Hill Tel: 1-919-962-5696 http://miaolab.org<http://miaolab.org/> Editor-in-Chief npj Drug Discovery https://www.nature.com/npjdrugdiscov On Jun 2, 2025, at 4:52 AM, Marc Ciruela Jardí via GaMD-discuss <gam...@li...> wrote: Dear GAMD developers, I have been trying to run GaMD on a membrane protein system with a small molecule and found several issues I would appreciate your help with. First, since I am trying to characterize the binding of the small molecule to the protein, should I start with the ligand bound to the protein or unbound? I don’t have a completely defined binding mode still, so it would be an approximation. The ligand binds in the interface of the protein with the membrane lipids, so in case I put it unbound, should be already in the membrane probably… Another doubt I have is what kind of biasing to apply. LiGaMD3 seems to fit the best, but because the membrane is also somehow involved in the binding, should I use another kind of GaMD? Maybe the dual boost GaMD where the boost is applied to all the system? In addition, I did some tests and have some doubts about the results. Following the indications in the tutorials, I did a GaMD equilibration (with a previously equilibrated system) with ntcmd = 1996360, ntcmdprep = 798544, nteb = 7985440, ntebprep = 798544 and ntave = 399272. I tried applying LiGaMD3 with the threshold in the upper bound (iEP=2, iED=2 but iEB=1) as indicated in the LiGaMD3 publication and with a series of sigma0P and sigma0D values, from 4 to 8 with sigma0B fixed at 6. To decide what values to use, I looked at the trajectories and observed that with sigma0P=sigma0D=4 I already found an unbinding event. Looking at the log, k0P already stabilized at 1 and k0D and k0B were near 1. The values of the last unboosted potential energy and unboosted dihedral energy where 30.238509140909 and -226779.050457056845. The dihedral energy is very small, right? Why may this be happening? I continued with a production run but never found a rebinding event in 1us of simulation. Despite this, I decided to try and analyse the simulation to calculate the PMF. The dV distribution was centered around 30-25 kcal, which seems quite high, but I suppose it’s because of iE=2. The distribution seemed normal, but a bit skewed. I tried doing a reweightening using several reaction coordinates but the profiles where very sharp, I suppose because of the lack of re-binding processes which gives a bad sampling. With the very low energetic values of the dihedral energy and high dV, I decided to try iE=1 but then, with the same sigma values I did not obtain any unbinding during the equilibration and the values for the potentials remained similar during the equilibration. Finally I tried with igamd=3 setting iEP=iED=1 and sigma0P=sigma0D=6. Here, I also didn’t see any unbinding but interestingly, the last unboosted potential energy and unboosted dihedral energy in the equilibration where -184058.981408115011 17220.120062542148, so completely different. Previously I had run a GaMD with a similar system with a peptide instead of a small molecule, using Pep-GaMD and starting from a quasi-bound state I was able to find a stable binding mode for the peptide. So I would much appreaicate your insight on what may be happening to this system and what recommendations would you have. Thank you for your time, Best regards, [cid:1cd9bdfb$1$1973954b49b$Coremail$jinan.wang$exeliris.ac.cn]<https://www.uab.cat/> Marc Ciruela Jardí Estudiant de doctorat Laboratori de Medicina Computacional Unitat de Bioestadística Edifici M - M3 · 08193 Bellaterra · Barcelona · Spain 93 581 38 12 www.uab.cat<https://www.uab.cat/> LMC<http://lmc.uab.cat/> [cid:5fc46ae$2$1973954b49b$Coremail$jinan.wang$exeliris.ac.cn]<https://orcid.org/0009-0001-9920-8789> Aquest missatge s'adreça exclusivament a la persona destinatària i pot contenir informació privada o confidencial. Si l'heu rebut per error, comuniqueu-nos-ho i destruïu-lo, i tingueu present que no teniu autorització per fer-ne cap ús. Aquest missatge no s’ha de respondre fora de la jornada laboral, en compliment de l’article 88 de la Llei orgànica 3/2018 de protecció de dades personals i garantia dels drets digitals<https://www.boe.es/boe_catalan/dias/2018/12/06/pdfs/BOE-A-2018-16673-C.pdf>, que reconeix el dret a la desconnexió digital, i de la Normativa interna per preservar el dret a la desconnexió digital del personal de la Universitat Autònoma de Barcelona<https://www.uab.cat/doc/NormativaDesconnexioDigital>. Abans d'imprimir aquest missatge penseu en el medi ambient. This message may contain private or confidential information, and is addressed exclusively to its intended recipient. If you have received it in error, please notify the sender and delete it. Please bear in mind that you are not authorized to use it for any purposes. Please consider the environment before printing this email. _______________________________________________ GaMD-discuss mailing list GaM...@li...<mailto:GaM...@li...> https://lists.sourceforge.net/lists/listinfo/gamd-discuss |
