1: %% This document created by Scientific Word (R) Version 2.5
2: %% Starting shell: mathart1
3:
4:
5: \documentclass[12pt,thmsa]{article}
6: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
7: \usepackage{sw20aip}
8:
9: %TCIDATA{TCIstyle=article/art2.lat,aip,article}
10:
11: %TCIDATA{OutputFilter=LATEX.DLL}
12: %TCIDATA{Created=Mon Apr 23 17:39:33 2001}
13: %TCIDATA{LastRevised=Wed Jan 16 15:21:42 2002}
14: %TCIDATA{<META NAME="GraphicsSave" CONTENT="32">}
15: %TCIDATA{Language=American English}
16: %TCIDATA{CSTFile=article.cst}
17:
18: \input{tcilatex}
19:
20: \begin{document}
21:
22: \author{O. Sotolongo-Costa$^{1}$, F. Guzman$^{2}$, J. C. Antoranz$^{3}$, \and G. J.
23: Rodgers$^{4}$, O. Rodriguez$^{2}$, J.D.T. Arruda Neto$^{5}$, \and A. Deepman$%
24: ^{5}$ \and {\small 1.- Department of Theoretical Physics. U.H., Havana 10400,%
25: } \and {\small Cuba. \ \TEXTsymbol{<}oscarso@ff.oc.uh.cu\TEXTsymbol{>}} \and
26: {\small 2.- Institut of Nuclear Sciences and Technology, Havana 10600,} \and
27: {\small Cuba.\TEXTsymbol{<}guzman@info.isctn.edu.cu\TEXTsymbol{>}} \and
28: {\small 3.- Departamento de Fisica Matematica y Fluidos, UNED, 28040} \and
29: {\small Madrid. \TEXTsymbol{<}antoranz@apphys.uned.es\TEXTsymbol{>} } \and
30: {\small 4.-Department of Mathematical Sciences. Brunel University,} \and
31: {\small Uxbridge, UK.\TEXTsymbol{<}g.j.rodgers@brunel.ac.uk\TEXTsymbol{>}}
32: \and {\small 5.- Instituto de F\'{i}sica, Universidade de Sao Paulo, Brasil,
33: CEP:} \and {\small 05315-970 \TEXTsymbol{<}arruda@usp.br\TEXTsymbol{>}}}
34: \title{A non extensive approach for DNA\ breaking by ionizing radiation.}
35: \maketitle
36:
37: \begin{abstract}
38: Tsallis entropy and a maximum entropy principle allows to reproduce
39: experimental data of DNA double strand breaking by electron and neutron
40: radiation. Analytic results for the probability of finding a DNA\ segment of
41: length $l$ are obtained reproducing quite well the fragment distribution
42: function experimentally obtained.
43: \end{abstract}
44:
45: Atomic force microscopy (AFM) has revealed itself as an extremely useful
46: device in the analysis of very small structures and specially in DNA
47: fragment analysis as it has been shown in \cite{berman}.
48:
49: It is interesting to study the production of fragments in DNA as a result of
50: radiation, since the presence of radiation interacting with DNA\ molecules
51: can influence the properties of living cells up to a lethal extreme.
52:
53: On the other hand, DNA\ fragment analysis may help in the study of the
54: structural properties of genome texts, and then to understand general
55: principles of genetic sequences.
56:
57: The process of DNA double strand breaking was performed by irradiation of
58: DNA\ molecules with electrons and neutrons at different doses ( See \cite
59: {berman} ). Then, the length of the resulting fragments was measured. As a
60: result, the collection of fragments was found to obey a fragment size
61: distribution function (FSDF), which presents important characteristics from
62: the viewpoint of complexity.
63:
64: The main fact, which will be focused in this paper, is that the collection
65: of fragments is such that there is not a ``characteristic'' size of the
66: fragments, \textit{i.e}., the smaller the fragment, the more abundant is it.
67: The FSDF in this case does not present a definite local maximum, resembling
68: more to an inverse power law, i.e., a distribution function in the basin of
69: attraction of a stable (L\'{e}vy) distribution \cite{feller}.
70:
71: The main distinction of L\'{e}vy distributions lies in the fact that their
72: variance is divergent. Maybe because of it, scientists have paid attention
73: to them only recently.
74:
75: This feature of the FSDF is not new. It has been reported in \cite{ishii}
76: the occurrence of transition to scaling in FSDF during glass rods breaking,
77: in \cite{oddershede} the power law distribution of fragments was related to
78: self-organized criticality (SOC). Matsushita \cite{matsu} proposed a fractal
79: representation for a general process of fragmentation.
80:
81: Our group \cite{pre,prl} detected power law behavior in the process of
82: liquid drop fragmentation and we proposed a Bethe lattice representation to
83: interpret FSDF in these experiments.
84:
85: Some attempts to relate FSDF to first principles in physics like the maximum
86: entropy principle are present in \cite{englman,li-tankin} with results that,
87: at the best, do not cover the process in which scaling in FSDF\ is present. (%
88: \textit{i.e.}, when the energy of the fragmentation process is high).
89:
90: The universal nature and almost unlimited range of applicability of the
91: maximum entropy principle leads us to expect it to be useful in describing
92: scaling in FSDF even at DNA\ scale.
93:
94: But the process of fractionating, by its own nature, is a paradigm of
95: phenomena in which interactions are long-range correlated among all parts of
96: the object under fragmentation. Then, though the maximum entropy principle
97: is expected to have an unlimited range of application, in the process of
98: breaking the expression for the entropy in its Shannon form:
99:
100: \begin{equation}
101: S=-k\int p(x)\log p(x)dx \label{eq:1}
102: \end{equation}
103: -where $p(x)dx\label{eq:1}$is the probability of finding the system
104: magnitude $x$ in the interval $[x,x+dx]$ , and $k$ is Boltzmann's constant-
105: is not applicable.
106:
107: This is because this formula, based in Boltzmann-Gibbs statistics, is
108: expected to be valid when the effective microscopic interactions are
109: short-ranged, and this gives to this entropy its extensive character (The
110: entropy of the whole object equals the sum of the entropies of its
111: constituent independent parts).
112:
113: Since, as we already pointed out, all parts of the fractionating object
114: during the process of violent breakage are correlated, then the entropy of
115: the object being fractionated is smaller tan the sum of the entropies of the
116: parts in which the object divides, defining this way a ``superextensivity''
117: in this system. This suggests that it may be necessary to use non-extensive
118: statistics, instead of the Boltzmann-Gibbs one.
119:
120: This kind of theory has already been proposed by Tsallis \cite{tsallisjsp},
121: who postulated a generalized form of entropy, given by
122:
123: \begin{equation}
124: S_q=k\frac{1-\int p^q(x)dx}{q-1} \label{eq:2}
125: \end{equation}
126:
127: where q is a real number.
128:
129: This entropy can also be expressed as:
130:
131: \begin{equation}
132: S_q=-k\int p(x)l_qp(x) \label{eq:3}
133: \end{equation}
134:
135: where the generalized logarithm $l_qp(x)$ is defined as (See \cite
136: {tsallisbjp}):
137:
138: \begin{equation}
139: l_q(p)=\frac{p^{1-q}-1}{1-q} \label{eq:4}
140: \end{equation}
141:
142: It is straightforward to see that $S_q\longrightarrow S$ when $%
143: q\longrightarrow 1$ , recovering Boltzmann-Gibbs statistics.
144:
145: It is our goal to derive, starting from first principles, a functional
146: dependence to describe the DNA DSDF obtained in \cite{berman}.
147:
148: Starting from equation \ref{eq:2} we may follow the method of Lagrange
149: multipliers to apply the maximum entropy principle to the fragmentation of
150: DNA. To do this, we impose two constraints: The first is the trivial one of
151: normalization of the probability:
152:
153: \begin{equation}
154: \int p(l)dl=1 \label{eq:5}
155: \end{equation}
156:
157: \textit{i.e., }the sum of the probabilities of finding a fragment of any
158: length is equal to unity.
159:
160: As a second constraint we may choose to adopt a ``q-mean value'' as:
161:
162: \begin{equation}
163: \int p^q(l)ldl=1 \label{eq:6}
164: \end{equation}
165:
166: Which reduces to the classical mean value when $q\longrightarrow 1$ .In this
167: formulation the length $l$ of the fragments has been referred to a unit
168: adequately chosen as to choose the ''q-mean value'' equal to one.
169:
170: It may seem strange to introduce a ``q-mean value'' , also known as
171: ``unnormalized mean value'' in this formulation. Really, this choice is not
172: unique but for our purposes and for simplicity reasons we will choose this
173: formulation. In \cite{tsallisbjp} a detailed discussion of the possible
174: choices for the second constraint can be found. The one here chosen showed
175: to be particularly useful in describing anomalous diffusion and was also
176: employed by us in \cite{entropy,physA} dealing with problems of
177: fragmentation.
178:
179: Now we use the method of Lagrange multipliers by means of the construction
180: of the functional:
181:
182: \begin{equation}
183: \pounds (p_i,\alpha ,\beta )=S_q-\alpha \int p(l)dl+\beta \int p^q(l)ldl
184: \label{eq:7}
185: \end{equation}
186:
187: being $\alpha $ and $\beta $ the Lagrange multipliers.
188:
189: The extremization of this functional leads to:
190:
191: \begin{equation}
192: p(l_i)=\frac{\beta (2-q)dl}{[1+\beta (q-1)l]^{1/(q-1)}}; \label{eq:8}
193: \end{equation}
194:
195: Alternatively, the same method when applied to the Boltzmann entropy ($q=1$)
196: gives
197:
198: \begin{equation}
199: p(l)dl=\beta e^{-\beta l}dl. \label{eq.9}
200: \end{equation}
201: \label{eq.9}Equation \ref{eq:8} is the expression for the probability of
202: finding a fragment of length $l_i$ and depends on three coefficients to
203: adjust. In this case we can apply this expression to fit it to the
204: experimental data reported in \cite{berman}., where methods of atomic force
205: microscopy were applied to measure FSDF of irradiated DNA.
206:
207: Figure 1 shows the experimental results for DNA breaking with electrons at
208: doses of 5000 and 7000 Gy. Both are fitted with Eq. \ref{eq:8}. Figure 2
209: represents FSDF for DNA breaking with neutrons at the same doses. In both
210: cases the length of the fragments was normalized to the length of the
211: largest one, and the number of fragments was normalized to the total number
212: of fragments. As it can be seen, the agreement is very good.
213:
214: More experimental data for electrons from 50 to 200 Gy and neutrons at doses
215: of 900, 7500, 2000 and 10000 Gy were also fitted with good results. In this
216: paper we are reporting the results of the coincident doses of electrons and
217: neutrons of 5000 and 7000 Gy to illustrate the application of this
218: viewpoint. Only in the cases of very low doses of electrons (50 and 100 Gy,
219: where fluctuations in FSDF are important) the results are not as good as the
220: ones before.
221:
222: This fact reveals the non extensive nature of DNA breaking, as it was shown
223: for macroscopic objects in \cite{entropy,physA}. So, This characteristic of
224: breaking is not exclusive of macroscopic bodies.
225:
226: Use of Boltzmann\'{}s entropy to describe FSDF obtained in these experiments
227: leads to incorrect results, (\textit{i.e}.,\ref{eq.9}) impossible to fit
228: with the data, which shows power law behavior.
229:
230: On the other hand, this non extensivity may also reflect an intrinsic nature
231: of the very DNA chain. The presence of $1/f$ spectrum in sequences and
232: long-range correlations in the DNA\ sequences \cite{voos} supports this
233: assertion.
234:
235: This work has been partially supported by the ``Alma Mater'' contest, Havana
236: University. One of us (O.S.) is indebted to the Department of Mathematical
237: Physics and Fluids, UNED, for kind hospitality.
238:
239: \begin{thebibliography}{99}
240: \bibitem{berman} D. Peng \textit{Atomic Force Microscopy investigation of
241: electron and neutron induced DNA double strand breaks. }Ph. D. Thesis. The
242: George Washington University, Washington D.C. 1997.
243:
244: \bibitem{feller} Feller, W. \textit{An Introduction to Probability Theory
245: and Applications} Vol. 2 Wiley, New York, 1966.
246:
247: \bibitem{ishii} T. Ishii and M. Matsushita, J. Phys. Soc. Jpn. \textbf{61},
248: 3474 (1992).
249:
250: \bibitem{oddershede} L. Oddershede, P. Dimon and J. Bohr, Phys. Rev. Lett.%
251: \textbf{71}, 3107 (1993).
252:
253: \bibitem{matsu} M. Matsushita J. Phys. Soc. Jpn. \textbf{54}, 857 (1985).
254:
255: \bibitem{pre} O. Sotolongo-Costa, E, Lopez, F. Barreras, J. Marin Phys.
256: Rev. E. \textbf{49}, 4027 (1994).
257:
258: \bibitem{prl} O. Sotolongo-Costa, Y. Moreno, J. Lloveras and J. C. Antoranz
259: Phys. Rev. Lett. \textbf{76}, 42 (1996).
260:
261: \bibitem{englman} R. Englman, N. Rivier and Z. Jaeger Phil. Mag.\textbf{65}%
262: , 751 (1987).
263:
264: \bibitem{li-tankin} X. Li, R. S. Tankin Combust. Sci. and Tech. \textbf{56}%
265: , 65 (1987).
266:
267: \bibitem{tsallisjsp} C. Tsallis J. Stat. Phys. \textbf{52}, 479 (1988).
268:
269: \bibitem{tsallisbjp} C. Tsallis Brazilian J. Phys. \textbf{29}, 1 (1999).
270:
271: \bibitem{entropy} O. Sotolongo-Costa, Arezky H. Rodriguez and G. J. Rodgers
272: Entropy.\textbf{\ 2}, 172 (2000) (http://www.mdpi.org/entropy)
273:
274: \bibitem{physA} O. Sotolongo-Costa, Arezky H. Rodriguez and G. J. Rodgers
275: Physica A \textbf{286}, 210-214 (2000).
276:
277: \bibitem{voos} R. F. Voos Phys. Rev. Lett. \textbf{68}, 3805 (1992). See
278: also M. de Sousa Vieira cond-matt/9905074.
279: \end{thebibliography}
280:
281: \subsection{Figure Captions}
282:
283: Fig.1: Normalized DSDF for electron irradiation of DNA at 5000 Gy and 7000
284: Gy. The solid squares represent the experimental results at 7000 Gy and
285: solid circles at 5000 Gy. Fitting was made with Equation \ref{eq:8}. Solid
286: curve is for 5000 Gy. Curve with open squares is for 7000 Gy.
287:
288: Fig. 2: Normalized DSDF for neutron irradiation of DNA at 5000 Gy and 7000
289: Gy. The solid squares represent the experimental results at 7000 Gy and
290: solid circles at 5000 Gy. Fitting was made with Equation \ref{eq:8}. Solid
291: curve is for 5000 Gy. Curve with open squares is for 7000 Gy.
292:
293: \end{document}
294: