Re: CCL:MD & MC; Modelling of counterions effects in polyelectrolytes.

 	I would like to comment on the MD simulations of the DNA. We have
 spent the past six years trying various methods to simulate the DNA with
 and without counterions. As you correctly point out, it is a serious
 headache, especially while trying to do a nanosecond level MD on DNA. Here
 are some of our observations (We use WESDYN, a derivative of GROMOS, and
 the force field used is GROMOS 86 with SPC water. All discussions here
 refer to the dodecamer sequence CGCGAATTCGCG):
 	1. Our goal is to simulate a "dilute aqueous solution" of DNA.
 As a result, we do not apply solute-solute boundary conditions - one
 should configure the box/cutoff correctly to avoid some of the artifacts
 of arising from this model.
 	2. In order for 1 to work correctly, we cannot have the
 counterions roam freely, so they are confined to a region around the
 surface of the DNA using a harmonic restraint. (In some sense, if we did
 this to 75% of the counterions, we will be consistent with the Manning
 counterion-condensation model!)
 	3. With 1 and 2, the DNA typically looks "OK" (by conventional
 convergence criteria) for the first 150ps or so, and then all sorts of
 things happen. Basically, the harmonic restraint shifts the ions to be
 between two Phosphate groups and the DNA begins to overwind itself etc.
 	4. So, we decided to skip the ions altogether. (We, in parallel,
 began a NaCl solution simulation to understand the effects of various
 protocols on the outcome of the simulation and found really strange
 things, such as the like-ions showing an enormous peak in g(R) at whatever
 the non-bonded cutoff is. This is likely to happen in the DNA simulation
 also. This work was done by Pascal Auffinger and is being currently
 written up) We scaled all Phosphate group charges down to -.25 by simply
 multiplying the original charges by .25. (CHARMM has a scaled charge
 model, where the total Phosphate group charge adds up to -.25, but the
 individual atomic charges are still very high)
 	5. We have a nanosecond MD with this model, showing very stable
 DNA structure. A preliminary communication of this appeared in JACS, vol.
 116, 4461-4462 (Authors Kevin McConnell A detailed paper is under
 	We have had reasonable success with this model for various other
 sequences ( as measured by the DNA structural stability, and calculated
 NMR properties compared against measured properties) also. We are
 currently rethinking the whole protocol and we will soon be getting back
 to the explicit counterions in some form or the other.
 	It also is a fact that for non-canonical B-DNA, the net charges on
 phosphates need not be -.24 (see the paragraph below), so anyone wanting
 to model Manning's theory, should be constantly readjusting the fractional
 charges on the phosphates (may not be much of a change!!!!).
 	I offer my sincere apologies, if the following paragraph is a
 restatement of a well-known thing and therefore a waste of bandwidth:
 	The net charge of -.25 on phosphate groups is valid ONLY for
 B-DNA at 298 K. The general condensation formula is:
 	f = 1./ (N * Zeta)		Zeta > 1./N
 	  = 1.				Zeta < 1./N
 	Here f in the uncompensated charge fraction on the phosphate, N is
 the valence of the counterion and Zeta is a dimensionless measure of
 polyion charge density.
 	Zeta = q**2 / (epsilon*k*T*b)
 	q is the protonic charge, epsilon, the dielectric constant of
 solvent, k , the Boltozmann constant, T, the temperature and b is the
 average linear charge spacing between the phosphates along the helical
 	For an aqueous solution of canonical B-DNA at 298 K, the
 fractional charge is -.24 for Na+ and -.12 for Mg2+.
 * Ganesan Ravishanker			Ph: (203) 344-8544 Ext. 3110       *
 * Coordinator of Scientific Computing,  Fax:(203) 344-7960                 *
 * Adjunct Associate Professor(Dept. of Chem.)                              *
 * Wesleyan University               e-mail:ravishan { *at * }  *
 * Middletown, CT 06459.                                                    *