1: \documentclass[a4paper,12pt]{article}
2: %\documentclass[a4paper]{article}
3: %\documentclass[twocolumn,a4paper]{revtex4}
4: \usepackage[all]{xy}
5: \usepackage{amsfonts}
6: \usepackage{amssymb}
7: \usepackage{epsfig}
8: \usepackage{graphicx}
9: \usepackage{citesort}
10:
11: %\usepackage{showkeys}
12:
13: %PARA PONER ACENTOS.....................
14: %\usepackage[spanish]{babel}
15: %\usepackage[T1]{fontenc}
16: \usepackage[latin1]{inputenc}
17: %PARA PONER ACENTOS......................
18:
19: \def\beq{\begin{equation}}
20: \def\eeq{\end{equation}}
21: \def\bea{\begin{eqnarray}}
22: \def\eea{\end{eqnarray}}
23: \def\winf{W_{1+\infty}\ }
24: \def\u1{\widehat{U(1)}}
25: \def\su2{\widehat{SU(2)}_1}
26: \def\suem{\widehat{SU(m)}_1}
27: \def\rep{representation }
28: \def\reps{representations }
29: \def\scr{\scriptstyle}
30: \def\disp{\displaystyle}
31: \def\und{\underline}
32: \def\I{{\rm Im}\ }
33: \def\R{{\rm Re}\ }
34: \def\Z{{\bf Z}}
35: \def\nl{\nonumber \\}
36: \def\ri{\rm i}
37: \def\re{\rm e}
38: \def\a{\alpha}
39: \def\b{\beta}
40: \def\g{\gamma}
41: \def\d{\delta}
42: \def\e{\eta}
43: \def\k{\kappa}
44: \def\l{\lambda}
45: \def\t{\tau}
46: \def\de{\partial}
47: \def\eps{\varepsilon}
48: \def\ov{\overline}
49:
50: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
51: \catcode`\@=11
52: \def\numberbysection{\@addtoreset{equation}{section}
53: \def\theequation{\thesection.\arabic{equation}}}
54: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
55:
56: %\def\baselinestretch{1.2}
57: \parskip 6pt
58: \parindent 14pt
59: \voffset -1cm
60: %\hoffset -0.5cm
61: \textwidth 15cm \textheight 22cm
62: %\oddsidemargin -0.5cm
63: %\evensidemargin -0.5cm
64:
65: %\numberbysection
66:
67: \begin{document}
68:
69:
70: \begin{titlepage}
71: \begin{center}
72: \hfill \quad q-bio/yymmnnn \\
73: \vskip .5 in {\Large\bf Thermodynamics of a model}
74:
75: {\Large\bf for RNA folding} \vskip 0.3in
76:
77: Mat\'{\i}as G. dell'Erba\\
78: {\em Departamento de F\'{\i}sica, Facultad de Ciencias Exactas y
79: Naturales, Universidad Nacional de Mar del Plata, Funes 3350,
80: (7600), Mar del Plata, Argentina}
81: \\
82: \vskip 0.3in Guillermo~R.~Zemba\footnote{ Member of CONICET,
83: Argentina.}
84: \\
85: {\em
86: Facultad de Ciencias Fisicomatem\'aticas e Ingenier\'{\i}a,
87: Universidad Cat\'olica Argentina, Av. A. Moreau de Justo 1500, Buenos Aires, Argentina}
88: \\
89: and
90: \\
91: {\em Departamento de F{\i}sica, C.N.E.A.
92: % Comisi\'on Nacional de Energ\'{\i}a At\'omica,
93: Av.Libertador 8250, (1429) Buenos Aires, Argentina}
94: \end{center}
95: \vskip .5 in
96: \begin{abstract}
97: \noindent We analyze the thermodynamic properties of a simplified
98: model for folded RNA molecules recently studied by G. Vernizzi, H.
99: Orland, A. Zee (in {\it Phys. Rev. Lett.} {\bf 94} (2005) 168103).
100: The model consists of a chain of one-flavor base molecules with a
101: flexible backbone and all possible pairing interactions equally
102: allowed. The spatial pseudoknot structure of the model can be
103: efficiently studied by introducing a $N \times N$ hermitian random
104: matrix model at each chain site, and associating Feynman diagrams
105: of these models to spatial configurations of the molecules. We
106: obtain an exact expression for the topological expansion of the
107: partition function of the system. We calculate exact and
108: asymptotic expressions for the free energy, specific heat,
109: entanglement and chemical potential and study their behavior as a
110: function of temperature. Our results are consistent with the
111: interpretation of $1/N$ as being a measure of the concentration of
112: $\rm{Mg}^{++}$ in solution.
113: \end{abstract}
114:
115: %\bigskip
116:
117: PACS numbers: 87.14gn , 02.10.Yn , 87.15.Cc
118:
119: \vfill
120: \end{titlepage}
121: \pagenumbering{arabic}
122:
123:
124: %-1------------------------------------
125:
126:
127: The applications of mathematical and statistical mechanics
128: techniques to study suitable biological problems has been a
129: successful area of recent research interest \cite{Gen1,Gen2,Gen3}.
130: In particular, the study of the spatial and topological
131: (pseudoknot) structure of DNA and RNA molecules is a successful
132: example of the above
133: \cite{Muller,Zee1,Zee2,Zee3,Zee4,Zee5,David,Tinoco}. A RNA
134: molecule is a heteropolymer strand made up of four types of
135: nucleotides: uracil ($U$), adenine ($A$), guanine ($G$), and
136: cytosine ($C$). The sequence of these bases from the $5'$ to the
137: $3'$ end defines the primary structure of the molecule. In
138: solution, at room temperature, different bases can pair with each
139: other by means of saturating hydrogen bonds to give the molecule a
140: stable shape in three dimensions, with $U$ bonding to $A$, $C$ to
141: $G$, and wobble pair $G$ to $U$, all with different interaction
142: energies. This last interaction (non Watson-Crick base pair)
143: together with triplets, quartets, etc. has a important role in
144: fold of the RNA molecule \cite{Tertiary,Multi,GU,Zhang}. The
145: effect of stacking interactions also contributes to the stability
146: of the molecule, making sets of adjacent bonds twist into the
147: familiar Watson-Crick helices. Among all possible structures that
148: arise from interaction between the bases, one defines the
149: secondary structures of a RNA molecule as all structures which are
150: represented by planar arc diagrams, that is, no crossing of arcs
151: in a representation resembling a Feynman diagram. When the
152: diagrams are non-planar, one says that a RNA molecule contains one
153: or more pseudoknots (see \cite{knots1,knots2} for a general
154: definition and \cite{Zee6} for a discussion on
155: the planarity of the diagrams). Finally, one defines the tertiary
156: structure of RNA as the actual spatial three-dimensional
157: arrangement of the base sequence.
158:
159: Several methods have been successfully used to study the folding
160: dynamics of RNA molecules in various conditions. Some of these are
161: based on statistical mechanics models, which usually avoid the
162: complexities related to the dynamical evolution of the real world
163: molecules, but allowing for a simple, kinematical treatment of the
164: proposed models. Therefore, the study of these models can shed
165: light on the intricate molecular dynamics and is our main
166: motivation for considering them. In this paper, we study a
167: simplified model of a RNA-like molecule considered in \cite{Zee1}
168: in which the geometric degrees of freedom of the system, such as
169: the stiffness and the sterical constraints of the chain, are not
170: taken into account. In addition, they consider that all pairs of
171: bases interact with a common pairing strength
172: (the assumption of neglecting disorder along the sequence is
173: actually a classic approximation \cite{deGennes}).
174: Moreover, the model keeps the fundamental
175: property of saturation of the interactions, that is, given a base
176: in the chain it can interact (following the rules mentioned above)
177: with only one another base at a time. The study of this model is
178: interesting in itself and has motivated some interesting work in
179: the literature, including case of the planar diagram limit
180: (no pseudo-knots) \cite{model}\cite{model2}\cite{model3}, and
181: the study of the tertiary structure of
182: the RNA molecule (see, for instance
183: \cite{Zee1,Zee2,Zee3,Zee4,Zee5}).
184: Moreover, the model allows for
185: some exact calculations including the partition function, the
186: specific heat, and some other thermodynamical and physical
187: quantities.Therefore, the study of the physical properties of this
188: model could be considered as a first approximation or a limit case
189: for more realistic RNA models. A natural extension of the model in
190: \cite{Zee1} towards more realistic ones (for example, including
191: different interactions between the bases) could be developed by
192: simple modifications of a matrix potential.
193:
194: The authors of \cite{Zee1} consider a system of $L$ molecules
195: (nucleotide bases) forming a lineal macromolecule with the shape
196: of a chain. They do not describe the formation of the backbone,
197: but only the interaction between links of the chain that produce
198: the folding of the RNA macromolecule (see Fig.\ref{Plegamiento}).
199: Each base can interact through an attractive force with any other
200: base of the chain. However, once a given molecule has paired with
201: other, it will not interact again with another. In this case, it
202: is said that the interaction between these bases saturate.
203:
204:
205: %-------Fig 1 Plegamiento -------------------------------------
206: \begin{figure}[h]
207: %\epsfysize=12cm \centerline{\epsfbox{Plegamiento.eps}}
208: \includegraphics[trim=0mm 43mm 17mm 10mm,clip,width=12cm]{Plegamiento.eps}
209: \caption{Arc diagram representation of the interacting pairs
210: $(1,3)$ and $(2,6)$ and its respective folded diagram.}
211: \label{Plegamiento}
212: \end{figure}
213: %-------Fig 1 Plegamiento -------------------------------------
214:
215: Although the bases that form the RNA molecule interact with
216: different pairwise energies, a first simple approximation is to
217: consider that this energy is the same for all bonds, and that any
218: paring of bases is assumed to be feasible. This amounts to
219: considering just one type of base and no further selection rules.
220: Note that if all energies are equal, then the Boltzmann factors
221: $v=e^{-\frac{\epsilon}{\kappa T}}$ (where $T$ is the absolute
222: temperature and $\kappa$ is the Boltzmann constant which we will
223: equal to one) are equal as well.
224: The configurational partition function of a molecule of
225: size $L$ in the model in \cite{Zee1} can be written as
226:
227: \beq
228: Z(L,N,T)=\frac{1}{N}\langle{\rm
229: tr}(1+\frac{1}{L^{1/2}}\varphi)^{L}\rangle=\frac{\int d\varphi\
230: e^{-\frac{N}{2Lv}{\rm tr}\varphi^{2}}\frac{1}{N}{\rm
231: tr}(1+\frac{1}{L^{1/2}}\varphi)^{L}}{\int d\varphi\
232: e^{-\frac{N}{2Lv}{\rm tr}\varphi^{2}}}, \label{Zvalormedio}
233: \eeq
234: \noindent
235: where $\varphi$ is a random $N \times N$ hermitian matrix and $Z$
236: depend of $T$ through $v$. Note that the simple form of
237: (\ref{Zvalormedio}) is only a consequence of the symmetry of the
238: matrix potential that reduces the original integration over $L$
239: matrices to one integration over $\varphi$ \cite{Zee1}. From the
240: theory of random matrices (see for example \cite{Mehta}, pag.
241: 140-2) it follows that
242: \beq
243: \langle {\rm tr}\varphi^{2k}
244: \rangle=\frac{(2k)!}{k!}\left(\frac{Lv}{2N}\right)^{k}
245: \sum_{j=0}^{k}
246: \left(%
247: \begin{array}{c}
248: k \\
249: j \\
250: \end{array}%
251: \right)
252: \left(%
253: \begin{array}{c}
254: N \\
255: j+1 \\
256: \end{array}%
257: \right)2^{j} \ ,
258: \eeq
259: \noindent where all averages are performed with respect to the
260: gaussian measure $d\varphi\ e^{-\frac{N}{2Lv} {\rm
261: tr}\varphi^{2}}$. Replacing this into (\ref{Zvalormedio}) and
262: taking into account that $\langle {\rm tr}\varphi^{k} \rangle=0$
263: for $k$ odd, we arrive at:
264: \beq
265: Z(L,N,T)=\sum_{k=0}^{[L/2]}\sum_{j=0}^{k}\left(%
266: \begin{array}{c}
267: L \\
268: 2k \\
269: \end{array}%
270: \right)
271: \left(%
272: \begin{array}{c}
273: k \\
274: j \\
275: \end{array}%
276: \right)
277: \left(%
278: \begin{array}{c}
279: N \\
280: j+1 \\
281: \end{array}%
282: \right) \frac{(2k)!}{2^{k-j}\ k!\ N^{k+1}}\ v^{k}
283: \label{PartFunct1}
284: \eeq
285: \noindent
286:
287: where the symbol $[L/2]$ means the integer part of $L/2$. From
288: (\ref{PartFunct1}) we may compute (for each $L$) $Z$ exactly, as a
289: function of $N$ and $T$. The large-$N$ asymptotic expansion of $Z$
290: has a well-known topological meaning \cite{'t Hooft}: the power of
291: $v$ is the number of arcs in the diagram, and the power of
292: $1/N^{2}$ is the genus $g$ of the diagram. It is therefore
293: convenient to write
294: (\ref{PartFunct1}) in following form:
295:
296: \beq
297: Z(L,N,T)=\sum_{k=0}^{[L/2]} d_{k}(L,N)\ e^{-\epsilon_{k}/T}
298: \label{PartFunctTrad}
299: \eeq
300: \noindent
301: where $\epsilon_{k}= k \epsilon$ and
302: \beq
303: d_{k}(L,N)=\sum_{j=0}^{k} \left(%
304: \begin{array}{c}
305: L \\
306: 2k \\
307: \end{array}%
308: \right) \left(%
309: \begin{array}{c}
310: k \\
311: j \\
312: \end{array}%
313: \right) \left(%
314: \begin{array}{c}
315: N \\
316: j+1 \\
317: \end{array}%
318: \right)\frac{(2k)!}{2^{k-j} \ k! \ N^{k+1}}
319: \label{Deg}
320: \eeq
321: \noindent
322:
323: From (\ref{PartFunctTrad}) we see that the spectrum of the system
324: has $[L/2]+1$ possible energies
325: $0,\epsilon,2\epsilon,\ldots,[L/2]\epsilon$ and the degeneracy of
326: the $k-th$ level is $d_{k}(L,1)$. For example, for $L=7$ the
327: maximum energy of a configuration is $3 \epsilon$ and there are
328: $d_{3}(7,1)=105$ different configurations with that energy.
329: Moreover, considering (\ref{Deg}) as a function of $N$ yields its
330: topological information, {\it e.g.}, for $L=7$, $d_{3}(7,N)=35+70\
331: 1/N^{2}$, which means that out of the total $105$ configurations
332: with $3 \epsilon$, $35$ have genus $0$ and $70$ have genus $1$.
333:
334: %-3.1---------------------------------
335: %\subsection{Asymptotic Limits of $Z_{L}$}
336:
337: Next, we calculate the partition function in the large $N$ limit,
338: which is the planar limit, using the results of \cite{Staudacher}.
339: For completeness, we quote here the results relevant for ours. We
340: define the resolvent $W(p)$:
341: \beq
342: W(p)=\frac{1}{N}\left \langle {\rm tr} \left(
343: \frac{1}{p-\varphi}\right)\right
344: \rangle=\sum_{n=0}^{\infty}\left(\frac{1}{p}\right)^{n+1}W_{n}\ ,
345: \eeq
346: \noindent
347: where $p$ is a complex variable, and
348: \beq
349: W_{n}=\frac{1}{N}\langle {\rm tr} \varphi^{n} \rangle \ .
350: \eeq
351: \noindent In the large $N$ limit, the resolvent is given by the
352: solution to the following equation called Pastur's equation
353: \cite{Pastur} (in the limit $g\rightarrow0$ and $c\rightarrow0$,
354: in \cite{Staudacher}):
355: \beq
356: W^{2}(p)-p \hspace{0.1cm} W(p)+1=0 \ ,
357: \eeq
358: \noindent which is:
359: \beq
360: W(p)=\sum_{k=0}^{\infty}C_{k}\frac{1}{p^{2k+1}}=\sum_{k=0}^{\infty}\frac{\left(%
361: \begin{array}{c}
362: 2k \\
363: k \\
364: \end{array}%
365: \right)}{k+1}\frac{1}{p^{2k+1}}\ , \label{Catalan}
366: \eeq
367: \noindent where $C_{k}$ are the Catalan numbers \cite{Gradshteyn}.
368: From (\ref{Catalan}) we obtain
369: \beq
370: \langle {\rm
371: tr}\varphi^{2k}\rangle=\frac{\left(%
372: \begin{array}{c}
373: 2k \\
374: k \\
375: \end{array}%
376: \right)}{k+1}L^{k}v^{k} \label{ValorMedioInf}
377: \eeq
378: \noindent With (\ref{ValorMedioInf}) we write down the partition
379: function in the large $N$ limit. We consider the case $N=1$ for
380: comparison purposes as well:
381: \beq
382: Z(L,N\rightarrow\infty,T)=\sum_{k=0}^{[L/2]} \frac{L!}{(L-2k)!\ k!\ (k+1)!}\ v^{k}
383: \eeq
384:
385: \beq
386: Z(L,N=1,T)=\sum_{k=0}^{[L/2]} \frac{L!}{(L-2k)!\ k!\ 2^{k}}\ v^{k}
387: \eeq
388: \noindent Note that both expressions for $Z$ are very similar,
389: except for the factors $(k+1)!$ and $2^{k}$ in the denominators of
390: the expansion coefficients. Noting that the first factor is larger
391: than the second, we conclude that $Z(N\rightarrow\infty) \leq
392: Z(N=1)$. The interpretation of this result is clear if we recall
393: that, for $v=1$, $Z(N\rightarrow\infty)$ counts the planar
394: diagrams only, whereas $Z(N=1)$ counts both the planar and
395: non-planar diagrams \cite{Zee1,Zee2,Zee3,Zee4}. Morevover, we
396: verify that both partitions functions coincide for values of $L$
397: smaller than $3$ as they should given that all diagrams are planar
398: in these cases.
399: Furthermore, in these two limiting cases, the partition function can be written as
400: in terms of hypergeometric functions:
401: \beq
402: Z(L,N=1,T)={_2F_{0}}(-\frac{L}{2},-\frac{L}{2}+\frac{1}{2};2v)\ ,
403: \label{hyper20}
404: \eeq
405: \beq
406: Z(L,N\rightarrow\infty,T)={_2F_{1}}(-\frac{L}{2},-\frac{L}{2}+\frac{1}{2};2;4v)\
407: , \label{hyper21}
408: \eeq
409: \noindent where
410: ${_pF_{q}}(\overrightarrow{a};\overrightarrow{c};z)=\sum_{k=0}^{\infty}\frac{(a_{1})_{k}\ldots(a_{p})_{k}}{(c_{1})_{k}\ldots(c_{q})_{k}}\frac{z^{k}}{k!}$
411: and $(a)_{k}=\frac{\Gamma(a+k)}{\Gamma(a)}$ are the $k$-order
412: Pochhammer symbols. We remark here that the results
413: (\ref{hyper20}) and (\ref{hyper21}) are exact.
414:
415: %-3.2---------------------------------
416: %\subsection{Topological Expansion of $Z_{L}$}
417:
418: As we mentioned before, the power of $1/N^{2}$ yields the genus
419: $g$ of the diagram, that is, the minimum number of handles of the
420: surface on which the diagram can be drawn without crossings. From
421: table of values of $Z$ for the smallest values of $L$ in
422: \cite{Zee1} we notice, for instance, that for $L=5$ the number of
423: planar diagrams on the sphere is $21$ and the number of non-planar
424: diagrams that can be drawn on a torus without crossings is $5$.
425: Next, we would like to write $Z(L,N,T)$ in the form of a
426: topological expansion \cite{'t Hooft,Zee1,Zee2}, that is, as a
427: power series in the variable $1/N^{2}$:
428: \beq
429: Z(L,N,T)=\sum_{g=0}^{\infty} z_{g}(L,T) \frac{1}{N^{2g}}\ ,
430: \label{ExpTop}
431: \eeq
432: \noindent where $z_{g}(L,\infty)$ is, for a molecule of size $L$,
433: the number of planar diagrams in a topological surface of genus
434: $g$. For the example of the previous paragraph, we have
435: $z_{0}(5,\infty)=21$ and $z_{1}(5,\infty)=5$. Note that
436: $z_{g}(L,T)$, as a function of $T$, is the partition function of
437: the system living on the topological surface of genus $g$. In
438: order to bring the partition function to the form (\ref{ExpTop}),
439: we first define the auxiliary function:
440:
441: \beq
442: G_{k}(N)=\sum_{j=0}^{k}
443: \left(%
444: \begin{array}{c}
445: k \\
446: j \\
447: \end{array}%
448: \right)
449: \left(%
450: \begin{array}{c}
451: N \\
452: j+1 \\
453: \end{array}%
454: \right)\frac{2^{j}}{N^{k+1}}. \label{FunAux}
455: \eeq
456: \noindent
457: This function contains all the $N$ dependence of
458: (\ref{PartFunct1}). Below, we write the binomial coefficient as
459: \beq
460: \left(%
461: \begin{array}{c}
462: N \\
463: j+1 \\
464: \end{array}%
465: \right)=\frac{1}{(j+1)!}N(N-1)\cdots(N-j)=\frac{1}{(j+1)!}\sum_{m=0}^{j+1}S_{j+1}^{(m)}N^{m}\ ,
466: \label{Stirling}
467: \eeq
468: \noindent
469: where $S_{j}^{(m)}$ is the Stirling number of the first
470: kind \cite{Gradshteyn,Stirling} with parameters $m,j$ (in turn, we
471: define $S_{j}^{(m)}=0$ if $m>j$ or if $j\leq0$). Replacing
472: (\ref{Stirling}) in (\ref{FunAux}), we obtain
473: \beq
474: G_{k}(N)=\sum_{j=0}^{k}\left(%
475: \begin{array}{c}
476: k \\
477: j \\
478: \end{array}%
479: \right)\frac{2^{j}}{(j+1)!}
480: \sum_{m=0}^{j+1}\frac{S_{j+1}^{(m)}}{N^{k+1-m}},
481: \eeq
482: Now, if we want to obtain the $O(1/N^{2g})$ of $G_{k}(N)$ (we
483: indicate this by $G_{k}^{(2g)}(N)$), we must require that
484: $k+1-m=2g$, then $j=k-2g,k-2g+1,\ldots,k$. To obtain all orders of
485: $N$ we must add all possible values of $g$:
486: \beq
487: G_{k}(N)=\sum_{g=0}^{\infty}G_{k}^{(2g)}(N)=
488: \sum_{g=0}^{\infty}\sum_{j=k-2g}^{k}\left(%
489: \begin{array}{c}
490: k \\
491: j \\
492: \end{array}%
493: \right)\frac{2^{j}}{(j+1)!}\frac{S_{j+1}^{(k+1-2g)}}{N^{2g}},
494: \eeq
495: replacing in (\ref{PartFunct1}) we obtain (\ref{ExpTop}) with
496: $z_{g}(L,T)$ given by: \beq
497: z_{g}(L,T)=\sum_{k=0}^{[L/2]}\sum_{j=k-2g}^{k}
498: \frac{L! \ 2^{j-k} \ S_{j+1}^{(k+1-2g)}}{(L-2k)! \ (k-j)! \ j! \
499: (j+1)!} \
500: e^{- \epsilon_{k}/T}. \label{ZxTop} \eeq
501: \noindent
502: In the limit $T\rightarrow\infty$, $z_{g}(L,T)$ coincides with $a_{L,g}$ of
503: \cite{Zee1}. Using the above mentioned property of the Stirling numbers,
504: we see from (\ref{ZxTop}) that the maximum
505: genus of a diagram for a given $L$ is $[L/4]$, therefore
506: $g\leq[L/4]$. An analysis of the $T$-dependent phase transition from
507: the topological expansion of the partition function will be given
508: in a future publication \cite{dellerba}.
509:
510: %-4---------------------------------
511:
512: %-4.1-------------------------
513:
514:
515:
516: In the rest of this letter, we analyze the thermodynamic
517: properties implied by the partition functions (\ref{PartFunct1})
518: by studying some interesting observables. We start by considering
519: the 'entanglement' (non-local two-point correlation function)
520: between two bases of the the chain in our model. For that, we use
521: the following definition for the correlation between the molecules
522: $i$ and $j$ ($i<j$) of the chain of size $L$
523:
524: \beq
525: \langle i,j \rangle=\frac{Z(\overline{j-i},N,T)}{Z(L,N,T)}, \label{corr-def}
526: \eeq \noindent
527: where $Z(\overline{j-i},N,T)$ is the partition function for
528: the molecule including bases $i$ up to $j$. For the case of
529: periodic boundary conditions we have
530: $Z(\overline{j-i},N,T)=Z(j-i+1,N,T)$. In the low temperature limit
531: the partition function becomes independent of $N$ (in this limit
532: only configurations up to two bases interacting are possible, it
533: is, planar configurations). Therefore yields the same result for
534: both the $N=1$ and $N\rightarrow\infty$ cases:
535:
536: \beq
537: \langle i,j
538: \rangle=\frac{1+\frac{1}{2}x(x+1)v}{1+\frac{1}{2}L(L-1)v}\simeq
539: 1-\frac{1}{2}(L^{2}-x^{2})e^{-\beta \epsilon}\ ,
540: \label{corr-vCERO}
541: \eeq
542: \noindent
543: with $L>>1$, $x=j-i>>1$ (provided $\beta
544: \epsilon>>\ln(L^{2}/12)+\ln(2+1/N^{2})$). The physical behavior of
545: this observable can be obtained by considering the situation where
546: $x$ is large, yielding $\langle i,j \rangle \simeq x^2 $, which
547: can be interpreted as a signal of confinement (long-range order
548: with critical exponent $-2$). This behavior is coherent with the
549: interpretation of the model as describing a folded RNA molecule.
550: Note that the exponent $(-2)$ for the long range order coincides
551: with the value found in \cite{model3,expo}.
552:
553:
554:
555: %-4.2----------------------------------
556:
557: Going ahead with our study of observables of the model, we now
558: calculate the normalized free energy $f$, or free energy per
559: molecule, in the limit of low temperature:
560: \beq
561: f(L,N,T)=\frac{F(L,N,T)}{L}=-\frac{1}{\beta L}\ln(Z(L,N,T))
562: \eeq
563: \noindent where $F$ is the free energy of the system. For $T<<1$,
564: both the $N=1$ and $N\rightarrow\infty$ cases yield:
565: \beq
566: f(\beta>>1)=-\frac{1}{2}(L-1)\frac{e^{-\beta \epsilon}}{\beta}
567: \eeq
568: \noindent
569: Furthermore, for the special case of $v=\frac{1}{4}$ (for which
570: $\beta=\ln 4/\epsilon$) we obtain
571: \beq
572: f(N\rightarrow\infty,v=1/4)=-\frac{1}{\beta}\ln(2)
573: \eeq
574:
575: %-4.3---------------------------------
576:
577: Next, we define the chemical potential of the model as:
578:
579: \beq
580: \mu(L,N,T) =\frac{\partial F(L,N,T)}{\partial N} =
581: -\frac{T}{Z(L,N,T)}\frac{\partial Z(L,N,T)}{\partial N}
582: \eeq
583: \noindent
584: The interpretation of $\mu$ is the following: we consider that
585: there is a gas of $N$ 'particles' in the internal space of the
586: random matrices at each site of the chain of size $L$. The
587: chemical potential measures the response of the system to a change
588: in the size $N$ of the matrix $\varphi$. On the other hand, the
589: concentrations of secondary and tertiary structures can be
590: separated experimentally by varying the concentration of
591: $\rm{Mg^{++}}$ ions in solution \cite{Mg++,Zee2,Zee5}; in the
592: original model \cite{Zee1}, one can assign this role of regulation
593: to $N$, as it is mentioned in \cite{Zee5} and can be seen from
594: (\ref{ExpTop}) that this dependency is how $1/N \sim
595: [\rm{Mg^{++}}]$. Therefore, the chemical potential $\mu$ can be
596: considered as a measure of the influence of the concentration of
597: $\rm{Mg^{++}}$ ions in solution on the system. From (\ref{ExpTop})
598: we see that, in the large $N$ limit, $Z$ is $O(1)$ and $\partial
599: Z/\partial N$ is $O(1/N^{3})$, therefore $\mu$ is $O(1/N^{3})$ in
600: the form:
601: \beq
602: \mu (L,N,T)= \frac{2 T}{N^{3}} \frac{z_{1}(L,T)}{z_{0}(L,T)} +
603: O(1/N^{5})
604: \eeq
605: \noindent
606: In the large $N$ limit, we obtain
607: the partition functions on the sphere and on the torus, $z_{0}$ and $z_{1}$
608: respectively, and write down the chemical potential
609: (see \cite{Stirling} for explicit expressions of $S_{n}^{(n-\alpha)}$)
610: :
611: \beq
612: \mu (L,N,T) \simeq \frac{T}{6 N^{3}} \langle k^{3}-k \rangle_{0},
613: \eeq
614: \noindent
615: where the averages are defined as:
616: \beq
617: \langle r_{k} \rangle_{0}
618: (L)=\left.\sum_{k=0}^{[L/2]}\frac{e^{-\epsilon_{k}/T}
619: r_{k}}{(L-2k)! \hspace{0.1cm} k! \hspace{0.1cm} (k+1)!
620: \hspace{0.1cm}} \right/
621: \sum_{k=0}^{[L/2]}\frac{e^{-\epsilon_{k}/T}}{(L-2k)!
622: \hspace{0.1cm} k! \hspace{0.1cm} (k+1)!},
623: \eeq
624: \noindent
625: and are labelled by the subindex $0$ in order to distinguish
626: them from the previously defined ones. Using numerical
627: calculations, it can be seen that for $T >> 1$, $\langle k^{3}\rangle_{0}>>
628: \langle k \rangle_{0}$ and $\langle k^{3}\rangle_{0}$ is
629: independent of $T$, then:
630: \beq
631: \mu (L,N,T>>1) \simeq \frac{\langle k^{3}\rangle_{0}}{6 N^{3}}
632: \hspace{0.05cm} T, \label{muT}
633: \eeq
634: \noindent
635: whereas for $T << 1$, we see that the dependence of $\mu$ on
636: $T$ is given by:
637: \beq
638: \mu (L,N,T<<1)\simeq \frac{(L-3)_4}{12 N^{3}} \hspace{0.05cm} T
639: e^{-2 \epsilon /T}. \label{mut}
640: \eeq
641: \noindent
642: The limits we have just discussed are summarized graphically
643: in fig. \ref{mu}. For any value
644: of the temperature, the system will tend to configurations with
645: large $N$, because this minimizes the chemical potential. In this
646: regime of $N$, the concentration of positive ions in solution
647: is small, and the configurations of the molecules
648: will tend to be planar.
649:
650: %-------Fig 2 Mu ----------------------------------------------
651: \begin{figure}[h]
652: \epsfysize=7cm \centerline{\epsfbox{mu.eps}} \caption{Chemical
653: potential as a function of the temperature for $N=100$.}
654: \label{mu}
655: \end{figure}
656: %-------Fig 2 Mu ----------------------------------------------
657:
658: The previous expressions for $\mu$ lead us to a consistent physical
659: interpretation of the parameter $N$. We recall that
660: \cite{Pathria}:
661:
662: \beq
663: \frac{\partial S}{\partial N}= - \frac{\mu}{T} ,
664: \eeq
665: \noindent
666: where $S$ is the entropy. From equations (\ref{muT}) and (\ref{mut}),
667: one can see that $S \propto 1/N^{2}$ in both $T \rightarrow 0$ and $T \rightarrow
668: \infty$ limits. Therefore, the entropy vanishes for $N \rightarrow
669: \infty$, which is also the limit in which the topology of the
670: molecule is spherical (by the topology of a
671: molecule, we mean topology of the Feynman diagrams associated
672: with the configuration of the molecule). This
673: suggests that $S$ could be considered as an indicator of
674: the spatial topological configurations of the
675: molecule. One can argue that the genus of the molecule is largely
676: determined by the conditions of the surrounding medium, such as
677: the concentration of $\rm{Mg^{++}}$ ions. The competition
678: between the interaction of a given base molecule with other
679: molecules in the chain and with the ions of the medium
680: regulates the
681: folding of the chain and therefore, its genus. One may
682: assume that the concentration of ions in the medium should be
683: monotonous functions of $1/N$. Therefore, we arrive to the conclusion
684: that the 'internal' parameter $N$, introduced by hand as a convenient
685: variable for organizing the topological configurations, could be given
686: the physical interpretation of representing the inverse quantity of
687: the ion concentration of the medium \cite{Zee2,Zee5,Zee6}.
688:
689: %-4.4----------------------------------
690:
691: Next, we consider the specific heat at
692: constant volume (in this case the volume is the size of the chain
693: $V=L$):
694:
695: \beq
696: C_{V}(T)=-T \left(\frac{\partial^{2}F}{\partial T^{2}}\right)_{L}
697: \label{Cv}
698: \eeq
699: \noindent
700: The graph of $C_{V}$ against the temperature (Fig. \ref{figCv}
701: (a)) has the particular shape characteristic of the system with
702: finite energy levels (see comment after equation (\ref{Deg})). The
703: characteristic temperature $T_{ch}$ corresponds to the
704: position of the peak in
705: $C_{V}$ showed in the graphs. For temperatures above and below
706: $T_{ch}$, the specific heat decreases rapidly. This well-known
707: behavior of the specific heat with temperature is known as the
708: Schottky anomaly \cite{Mandl,Pathria} and it is a general property
709: of systems with energy levels with discrete degeneracy (see
710: $d_k$ above).
711: In the low temperature region, we have
712: \beq
713: C_{V}(T<<1) \simeq k \left(\frac{\epsilon }{2k T}\right)^2
714: {\rm{sech}}^2\left(\frac{\epsilon }{2k T}-\ln
715: \sqrt{\frac{L(L-1)}{2}}\right) . \label{CvAprox}
716: \eeq
717: \noindent
718: In this limit, the specific heat coincides with that for a
719: two-level system, since for low enough temperatures, the system
720: will only be able to access the ground state and the first excited
721: state. In figure \ref{figCompCv} , we plot the exact specific
722: heat from (\ref{Cv}) and the low-temperature approximation from
723: (\ref{CvAprox}).
724: Furthermore, we may define the topological specific heat as:
725: \beq
726: C_{g}(T)=-T \left(\frac{\partial^{2}F_g}{\partial
727: T^{2}}\right)_{L, g} \label{Cvtop}
728: \eeq
729: where $F_g(L,T)=- \kappa T \ln(z_g (L,T))$. Note that $C_g$
730: can be identified with
731: the specific heat restricted to the diagrams on the topological
732: surface of genus $g$. In Fig. \ref{figCv} (b) we show that the
733: higher peaks correspond to the lower genera. This agrees with the
734: intuitive argument that considers a molecule with higher genus as
735: strongly folded and, therefore, with reduced
736: number of degrees of freedom. We can carry out
737: the same analysis for $C_{V}$, given that the addition of a new
738: base molecule increases the number of degrees of freedom of the system.
739: In this sense, we can consider the relation $L \sim 1/g$.
740:
741: %-------Fig 3 C_V y C_N ----------------------------------------
742: \begin{figure}[h]
743: \epsfysize= 5cm \centerline{\epsfbox{CvyCn.eps}} \caption{ a) $C_V
744: (T)$ for $N=10$ and $L=100, 200, 300$. b) $C_g (T)$ for $L=100$
745: and $g=0,10,20$.}\label{figCv}
746: \end{figure}
747: %-------Fig 3 C_V y C_N ----------------------------------------
748:
749: %-------Fig 4 Cv real vs Aproximado ------------------------------
750: \begin{figure}
751: \epsfysize=5.5cm \centerline{\epsfbox{RealvsAprox.eps}} \caption{
752: Exact and low-temperature approximation for the specific heat with
753: $L=20$ and $N=1$} \label{figCompCv}
754: \end{figure}
755: %-------Fig 4 Cv real vs Aproximado -------------------------------
756:
757: In conclusion,
758: we have studied several thermodynamical and topological aspects
759: of the simplified model of RNA of \cite{Zee1}.
760: We have presented an exact expression for the partition function
761: of the system, and gave an interpretation of the degeneracy of each
762: energy level as a function of $N$. Furthermore, we have calculated the
763: topological expansion of the partition function of the model, in
764: which the coefficients of the expansion can be interpreted as
765: the reduced partition functions for systems restricted to
766: topological surfaces of genus $g$. We showed that the
767: maximum genus of the configurations is $[L/4]$, for a molecule
768: of size $L$.
769: Moreover, we have calculated asymptotic expressions for some
770: thermodynamical observables, as a function of the temperature.
771: Analyzing the expressions for the chemical potential and
772: entropy, within our data, we find a consistent interpretation relating the variable
773: $1/N$ (arising from the matrix model) and the concentration of
774: $\rm{Mg}^{++}$, as it has
775: been suggested in \cite{Zee2,Zee5}.
776:
777: MdE thanks to Mat\'{\i}as Reynoso for helpful discussions.
778:
779: %------- Bibliography -------------------------------------
780: %
781: \def\NP{{\it Nucl. Phys.\ }}
782: \def\PRL{{\it Phys. Rev. Lett.\ }}
783: \def\PL{{\it Phys. Lett.\ }}
784: \def\PR{{\it Phys. Rev.\ }}
785: \def\CMP{{\it Comm. Math. Phys.\ }}
786: \def\IJMP{{\it Int. J. Mod. Phys.\ }}
787: \def\MPL{{\it Mod. Phys. Lett.\ }}
788: \def\RMP{{\it Rev. Mod. Phys.\ }}
789: \def\AP{{\it Ann. Phys.\ }}
790: \def\PRE{{\it Phys. Rev. E\ }}
791: \def\EPL{{\it Eur. Phys. Lett.\ }}
792:
793: \begin{thebibliography}{99}
794:
795: \bibitem{Gen1} V. S. Pande, A. Yu. Grosberg, T. Tanaka, \RMP {\bf 72} 1 (2000).
796: \bibitem{Gen2} I. M. Lifshitz, A. Yu. Grosberg, A. R. Khokhlov, \RMP {\bf 50} 3 (1978).
797: \bibitem{Gen3} E. Orlandini, S. G. Whittington, \RMP {\bf 79} 2 (1978).
798: \bibitem{Muller} M. M\"uller, \PRE {\bf 67} (2003) 021914.
799: \bibitem{Zee1} G. Vernizzi, H. Orland, A. Zee, \PRL {\bf 94} (2005) 168103.
800: \bibitem{Zee2} H. Orland, A. Zee, \NP B {\bf 620} (2002) 456-576.
801: \bibitem{Zee3} M. Bon, G. Vernizzi, H. Orland, A. Zee, \textit{J. Mol. Biol.} {\bf 379} (2008) 900-911, http://arXiv.org/q-bio/0607032.
802: \bibitem{Zee4} G. Vernizzi, P. Ribeca, H. Orland, A. Zee, \textit{Phys. Rev. E}, {\bf 73} (2006), 031902, http://arXiv.org/q-bio/0508042.
803: \bibitem{Zee5} M. Pillsbury, J. A. Taylor, H. Orland, A. Zee, \PRE {\bf 72} (2005) 011911, http://arXiv.org/cond-mat/0310505v2.
804: \bibitem{David} F. David, C. Hagendorf and K.-J. Wiese, \EPL {\bf 78} (2007) 68003.
805: \bibitem{Tinoco} I. Tinoco Jr. and C. Bustamante, \textit{J. Mol. Biol.} (1999) {\bf 293}, 271-281.
806: \bibitem{Tertiary} J. E. $\rm{\check{S}}$poner, K. Réblová, A. Mokdad, V. Sychrovský, J. Leszczynski and J. $\rm{\check{S}}$poner, \textit{J. Phys. Chem. B} (2007), {\bf 111}, 9153-9164.
807: \bibitem{GU} G. Varani and W. H. McClain, \textit{EMBO reports} (2000) {\bf 1}, 1, 18-23.
808: \bibitem{Multi} D.E. Gilbert and J. Feigon, \textit{Curr. Op. Struct. Biol.} (1999) {\bf 9}, 3, 305-314.
809: \bibitem{Zhang} Z. Li and Y. Zhang, \textit{Nucleic Acids Res.} (2005) {\bf 33}, 7.
810: \bibitem{knots1} D. W. Staple, S. E. Butcher, \textit{PLoS Biol} {\bf 3}(6), e213.
811: \bibitem{knots2} C. W. A. Pleij, K. Rietveld and L. Bosch, \textit{Nucleic Acids Res.} (1985) {\bf 13}, 5, 1717.
812: \bibitem{model} P.G. Higgs, {\it Phys. Rev. Lett.} {\bf 76} (1996) 704; R. Bundschuh and T.
813: Hwa, {\it Europhys. Lett.} {\bf 59} (2002) 903; A. Pagnani, G.
814: Parisi, and F. Ricci-Tersenghi, {\it Phys. Rev. Lett.} {\bf 84},
815: (2000) 2026; F. Krzakala, M. Mezard, and M. Müller, {\it Europhys.
816: Lett.} {\bf 57} (2002) 752.
817: \bibitem{model2} A. Montanari and M. Mezard, {\it Phys. Rev. Lett.} {\bf 86} (2001) 2178.
818: \bibitem{model3} M . Müller, {\it Phys. Rev.} {\bf E 67} (2003).
819: \bibitem{Zee6} G. Vernizzi, H. Orland, A. Zee, http://arXiv.org/q-bio/0405014v1.
820: \bibitem{deGennes} P. -G. de Gennes, Biopolymers {\bf 6} (1968) 715.
821: \bibitem{expo} M. Müller, F. Krzakala, and M. Mézard, {\it Eur. Phys. J.} {\bf E 9} (2002) 67-77.
822: \bibitem{Staudacher} M. Staudacher, Phys. Lett. {\bf B 305} (1993) 332-338.
823: \bibitem{Pastur} L.A. Pastur, Theor. Math. Phys. (USSR) {\bf 10},67 (1972).
824: \bibitem{Mehta} M. L. Mehta, \emph{Random Matrices}, Third Edition, Elsevier Academic Press (2004).
825: \bibitem{Gradshteyn} I. S. Gradshteyn, I. M. Ryzhik, A. Jeffrey, D. Zwillinger \emph{Table of Integrals, Series, and
826: Products}. Fifth Edition. Academic Press, New York (1994).
827: \bibitem{Stirling} M. $\rm{\check{Z}}$ivkovi$\rm{\acute{c}}$, Ser. Mat. Fiz. {\bf 498}-{\bf 541} (1975) 217-221.
828: \bibitem{Mandl} F. Mandl, \emph{Statistical Physics}, Second Edition, The Manchester Physics Series, UK (1988).
829: \bibitem{Pathria} R. K. Pathria, \emph{Statistical Mechanics}, Second Edition, Butterworth-Heinemann, UK (1996).
830: \bibitem{'t Hooft} G. 't Hooft, \NP B {\bf 72} (1974) 461.
831: \bibitem{dellerba} M. G. dell'Erba and G. R. Zemba (to appear).
832: \bibitem{Mg++} T. Gluick, R. Gerstner, D. Draper, {\it J. Mol. Biol.} {\bf 270}(1997) 451-463.
833: \\
834: \end{thebibliography}
835:
836:
837: \end{document}
838: