Skip to main content

Wondering about PME

   Well, enough of talk about non-scientific thinking. There is something about which I have wondered many times. The steps in doing any MD is as follows :

  1. Assign charges, atom type (sp3, sp2 etc with variations) and Van der Waals radii to each atom. Based on this, the parameters will be decided. As of now, most force fields use fixed charges which means their value does not change based on the environment. So, protonation states of His and others have to be determined empirically at this step.
  2. Following the parameter assignment to each atom, bond and torsion (together they constitute a force field), all short contacts etc need to be removed. Now, at this point, you place waters randomly around your protein and then 'energy minimize' your structure which relieves short contacts between waters and the protein. You can keep all atoms of the protein fixed, or allow some relaxation to take place at this step.
  3. Now, once the protein is ready (BTW, this is what is called a topology, although it is a stupid name for this file. I would rather have called it parameterized coordinate file or *.pcf), the atoms are all stationary as given in PDB file. Now, inside the cell, the atoms are never at rest, so we need to provide an initial guess for the velocities of each atoms. This is what is done at the time of equilibration. The initial guess is arrived at using the Boltzmann distribution so that the required temperature is obtained. This is also the first step in the equilibration process, and in Gromacs is known as the NVT simulation. The system (water + protein) is allowed in small steps to reach the designated temperature.
  4. After this, there is a pressure equilibration which maintains the correct number of atoms (both water and protein) within the volume we are simulating.
  5. The next step is where I have a problem. In this step, the simulation parameters are fed into the system and a run file is compiled. One thing that is done at this step is what is known as particle mesh ewald method. What you do is setup a cutoff distance below which you calculate the coulombic/VdW energy directly, and beyond the cutoff, the calculations are done in fourier space. Oneibig assumption for all this is to consider the MD system to be a periodic system (such as a crystal).
    Now we come to the part where I have a problem. Why do we consider the system to be periodic ? Inside the cell, the proteins are not arranged as in a crystal. So what is the use of such an assumption ? I agree that we need a continuous water system, but why should the protein also be repeated ? And it is usually because of such crystal system assumption that you need to add ions to neutralize the system. If that is not done, then unit cells will start repelling each other. But if we dont assume that, then first of all, we save a lot of time by not using the fourier space.summation. Secondly, we can use non-integer values for atom charges which by some accounts improve the simulations. So, as of now, I dont know why it is simulated as a crystal lattice. Maybe because the initial structure is a crystal structure. But that's not what we really want, so I am not so sure....

Comments

Post a Comment

Popular posts from this blog

Butter, Oil and Zephyr

    Well, have been doing a lot of a MD simulations lately. My current favourite is implicit solvent simulations. The entire explicit solvent thing always seemed too brute force to me. Its not elegant, and nor is it computationally feasible for larger systems. Implicit solvent simulations suffer from their own problems, but I would consider their future much brighter. This is simply sue to the larger time scales that you can access using these simulations even on a normal desktop system. Larger time scales means better averages and thermodynamic properties are always averages, so better match with experiment.    Speaking of Implicit solvent simulations, Gromacs does a good job, as well as any other currently. I especially like the OBC model for calculating Born radii which I am glad to say, I copied to MUST_READ folder when it first came out. It might still be lying there, that is another thing, but at least I can appreciate when a good...
1 comment
Read more

I am a Reverend now!!!

  I have been an atheist for most of my life. I have never truly believed that God exists or such thing. More recently, I have started looking down upon people that do believe in God. Not a hatred kind of looking down upon them, but rather a pitiful look down or maybe even a cuter look down upon. Like when you see a child earnestly insisting that Santa Claus exists. You pity the child's ignorance and you find it cute, but secretly you also get a bit tense knowing that his/her ignorance cannot last very long. And, you wonder if that earnest child could really handle the news when it hits them hard.   So, in summary all I am saying is, people who believe in a God are stupid. They should grow up and really think hard about what they mean when they say they believe in God. Half of them will mutter and sputter and fail to come up with what they really believe. Most would say something asinine like "I am sure there is something out there" or "There will be chaos if n...
Post a Comment
Read more

An early essay on co-translational folding

   One of the earliest essays I wrote on the co-translational folding concept. This was for a course on protein folding that Prof Balaram takes at IISc. I am reproducing it in full, with missing punctuation marks added. At one point, I say that I have not been able to find out why it is assumed that refolding is a good substitute for in vivo protein folding. I did know even then, but I chose not to talk about it. I did so, since I cannot talk about the topic without calling everyone involved an idiot. Also, I have omitted the references as I cannot bring myself to format them properly. Rest assured that nothing I said in the essay is without literature support. Co-translational folding Introduction    Protein folding is one of the major unsolved problems of molecular biology. The problem is as follows : given that a gene encodes a unique structure (which performs a unique function), how does this information conversion take place ? it is assumed (barrin...
Post a Comment
Read more