Dear Gromacs users, <br><br> I am new to essential dynamics, I have gone through<br>some fundamentals in PCS, the mailing list related to ED<br>and few publications by - <br>Amadei (Proteins, 17, 412-425, 1993),<br>a. Amadei (journal of biomolecular structure and dynamics, 13, 615, 1996)<br>
b. Berk Hess (Physical reviews E, 62, 8428-8448, 2000)<br>c. Berk Hess (Physical reviews E, 63, 031910, 2002)<br>d. Caterina (Biophysical chemistry, 92, 183-199, 2001)<br><br>I have made a short note of what I understand. Please correct.<br>
the mistakes.<br>1. ED is principally used to study the anharmonic motions, ie those <br> which are not equilibrated unlike equlibrated motions like bond<br> vibrations, bending etc.. (Equilibrated in the time scale of study)<br>
<br>2. So one has to run a simulation long enough so that it is more than <br> or equal to the time scale required for the specific motion (for eg.<br> closing of loop takes place in few ps, one has to run MD for atleast<br>
a ns to investigate it)<br><br>3. After MD for long enough time, when covariance matrix would indicate<br> whether two atoms move in same direction, or opposite over the time<br> of simulation based on positive or negative value. Extracting Eigenvalues<br>
and Eigenvectors from the matix gives the directions in which the highly<br> correlated motions occur. <br><br>4. Analysing all the data values projected over the first few eigen vectors<br> is the priciple component analysis. <br>
<br>5. if this experiment is done on same structure more than once (say 3), and the<br> first few priciple components of all 3 simulations coincide, then it could be <br> the most possible direction of motion in the protein otherwise the patern<br>
of PC's is most likely due to random motion.<br><br>My questions - <br>1. While chosing the period for covariance analysis, what is the criteria? in the <br> paper b, author mention that a certain peroid was chosen because the peptide <br>
free enegry minimum. Not clear about this, because when the protein resides in<br> an energy minimum how can there be transition to another configuration (eg a loop<br> movement) without crossing the barrier. should we not consider the time which <br>
spans a native conformation to say an active conformation during the simulation?<br><br>2. If we have done a long enough simulation of say 100ns for 4 proteins with similar <br> structure but with sequence id of 40-60% (different chain lengths 230-260aa), Can <br>
we do a covar analysis of these 4 simulations? <br><br>3. How much should be the socine content to tell that it is not a random diffusion?<br><br>4. Is ED analysis itself is not enough to establish a important movement in the <br>
protein.. further ED sampling is required to prove it?<br><br>5. Concept doubt - When all the structures at least square fitted before building a<br> covariance matrix, where is the random diffusion comming into picture? I am <br>
sorry I was not clear about this consept qualitatively.<br><br>6. I am working on a enzyme whcih upon binding to substrate shows a marginal<br> deviation from the native unbound structure. I have onlt the unbound form, but <br>
the two forms are available in other organisms whose structures are very similar<br> to the one I am working. So would you think doing the ED on this protein is <br> logical?<br><br>I was not recieving replies form the forum for my rescent queries (which I had to obtain<br>
from my senior, SO kindly check if there is any problem)<br><br>I AM EXTREMELY SORRY FOR VERY BIG MAIL.!!<br><br>Thanking you<br>With Regards<br>M. Kavyashree<br><br><br> <br><br>