1: \documentclass[aps,prl,twocolumn,groupedaddress]{revtex4}
2: \usepackage{graphicx}
3: \begin{document}
4:
5: \title{Kafri, Mukamel, and Peliti Reply}
6:
7: {\bf Kafri, Mukamel, and Peliti Reply}: The Comment of Hanke and
8: Metzler \cite{HM} questions the validity of the analysis presented
9: in \cite{KMP1} to DNA chains of finite length as used in
10: experiments. Their argument is that for the analysis to be valid
11: ``each of the three segments going out of a vertex must be much
12: longer than the persistence length $\ell_p$ of this segment''. By
13: using the persistence lengths $\ell_p(L) \sim 40$\AA~for a single
14: strand and $\ell_p(H) \sim 500$ \AA~for a double helix (bound
15: segment) they arrive at the conclusion that in order to observe
16: the asymptotic behavior found in \cite{KMP1} one needs chains
17: which are far longer than those studies experimentally.
18:
19: This assertion constitutes a misunderstanding of the analysis
20: given in \cite{KMP1}. In this analysis one considers a loop
21: interacting with {\it the rest of the chain} and not just with the
22: vicinal double helices. Thus, in \cite{KMP1} each of the two lines
23: attached to the loop is in fact composed of an alternating
24: sequence of bound segments and denaturated loops. It may be viewed
25: as a stick and joint structure, whereby adjacent double helices
26: (which may be considered as rigid rods as they are shorter than
27: $\ell_p(H)$) are loosely attached to each other via an open loop
28: (see Fig. 1). Thus, the ``rest of the chain'' as considered in
29: reference \cite{KMP1} is in fact a random rod structure with a
30: persistence length given by the length of the rod connecting two
31: loops. This length can be easily estimated from the analysis of
32: Poland and Scheraga and its generalization as given in
33: \cite{KMP1}. It can be shown that the probability distribution of
34: rods of length $k$ at the transition is $P(k) \sim (w/s)^k$. Here
35: $w=\exp(\beta_{\rm M} E_0)$ is the Boltzmann weight of a bound
36: pair with energy $-E_0$ at the melting temperature, $1/ \beta_{\rm
37: M}$, and $s$ is a non-universal geometrical factor. This is an
38: exponentially decaying distribution with a typical length of order
39: $\xi \sim 1/ \vert \ln(w/s) \vert$. This length has nothing to do
40: with the persistence length of the bounded segments and in fact it
41: is far shorter (of the order of a small number of base pairs). An
42: estimate of this length can be obtained using Eq. 4 of \cite{KMP1}
43: at the transition
44: \begin{equation}
45: \frac{s}{w}=\sum_{k=1}^{\infty} \frac{1}{k^c}+1.
46: \end{equation}
47: With the estimated value of $c=2.115$ (in three dimensions) this
48: yields $\xi \simeq 1.07$ in units of $\ell_p(L)$. Note that this
49: length is non-universal and it depends on the details of the
50: model, such as the statistical weight of small loops, the stacking
51: interactions and other details. However, these features are not
52: expected to change this length considerably. Therefore, the
53: criticism expressed in the Comment is not valid and the analysis
54: in the Letter is applicable to finite chains of the length studied
55: experimentally. In reference \cite{KMP1} the detailed structure of
56: the ``rest of the chain'' is ignored when the self-avoiding
57: interaction is considered. However, as argued in \cite{KMP1} this
58: assumption is reasonable and should yield a good estimate of the
59: effect of self-avoiding interactions.
60:
61: \begin{figure}
62: \includegraphics[width=1.5cm,angle=90]{DNAre.eps} \caption{Schematic
63: representation of a microscopic configuration of the DNA molecule.
64: Here the flexibility of the ``{\it rest of the chain}'' emerges
65: from the presence of loops which connect short but more rigid
66: bound segments.}
67: \end{figure}
68:
69: A strong support for this picture is provided by numerical
70: simulations of Causo {\it et. al.} \cite{CCG} and Carlon {\it et.
71: al.} \cite{COS} of chains of finite length in which the self
72: avoiding interactions are fully taken into account. In both
73: studies the melting transition was found to be first order,
74: compatible with $c>2$ as found in \cite{KMP1}. In \cite{COS} the
75: exponent $c$ is evaluated directly for chains of length 50-200
76: monomers (each monomer corresponds to $\ell_p(L)$ namely to about
77: 8 base pairs). Moreover, the inclusion of the different
78: stiffnesses of the bound and unbound segments (in simulations
79: persistence length ratio, $\ell_p(H) /\ell_p(L)$, was taken to
80: vary between 1 and 10) did not change the estimate significantly.
81: The fact that the observed $c$ is consistent with that obtained by
82: reference \cite{KMP1} strongly supports the basic assumptions
83: behind this analysis. It also indicates that the scaling behavior
84: predicted in \cite{KMP1} may be observed in finite chains of
85: lengths between hundreds to thousands base pairs, which is the
86: experimentally relevant range.
87:
88:
89: \noindent Y. Kafri and D. Mukamel
90:
91: Department of Physics of Complex Systems, The Weizmann Institute
92: of Science, Rehovot 76100, Israel
93:
94: \noindent L. Peliti
95:
96: Dipartimento di Scienze Fisiche and Unit\`a INFM, Universit\`a
97: ``Federico II'', Complesso Monte S. Angelo, I--80126 Napoli, Italy
98:
99:
100:
101: \begin{thebibliography}{99}
102:
103: %\input{bibliov1.tex}
104: \bibitem{HM} A. Hanke and R. Metzler cond-mat/0110164
105:
106: \bibitem{KMP1} Y. Kafri, D. Mukamel, and L. Peliti, Phys. Rev.
107: Lett. {\bf 85}, 4988 (2000); cond-mat/0108323.
108:
109: \bibitem{CCG} M. S. Causo, B. Coluzzi and
110: P. Grassberger, Phys. Rev. E {\bf 62}, 3958 (2000).
111:
112: \bibitem{COS} E. Carlon, E. Orlandini, and L. Stella, cond-mat/0108308.
113:
114: \end{thebibliography}
115:
116: \end{document}
117: