# A Novel Approach to Genome Editing using Cellular Automata Evolutions of Adjoints Sequences Abstract-This paper proposes a novel method for genome editing using cellular automata evolutions of adjoints of Adenine, Thymine, Guanine, and Cytosine. The adjoints of the given a genome sequence are the characteristic binary string sequences. For example, the adjoint of Adenine of a given genome sequence is a binary string consisting of 0's and 1's where 1's corresponds to the presence of Adenine in the genome sequence. So, one can have four adjoint sequences of Adenine, Thymine, Guanine, and Cytosine corresponding to a given genome sequence. One-dimensional three neighborhood binary value cellular automata rules can be applied to an adjoint sequence and the desired number of evolutions could be obtained. This rule is defined by a linear Boolean function and one can have 256 such linear Boolean functions. Genome editing is carried out by superimposing the evolved adjoint sequence on the original genome sequence or on its successive evolutions. In this manner, one can have four ways of genome editing using four adjoint sequences and evolutions. enome editing is essentially the process of introducing required changes in a given DNA. A protein or enzyme cuts certain portions of a given DNA and substitutes the target sequences of nucleotides by specific chains of nucleotides. For ages, people have been working on the notion of Genome editing. However, only recently, the spurt of activities has been reported in the literature about genome editing. Before getting into the details of the genome editing, it would be apt to reason out why genome editing is viewed as a significant activity. Genome editing is carried out as a health initiative to healeven the so-called incurable diseases associated with genetic problems. Clinically there are many methods of editing genomes among which CRISPR technique is recognized as a reliable technique ratified by the U.S. Food and Drug Administration, pushing in a new era of cancer treatment. CRISPR based therapy is designed to treat blood and bone marrow cancer, which usually affects children and young adults. This therapy is known G as CAR-T therapy, and it has shown remarkable results in patients. By eliminating those genes which cause disease, physicians can treat various illnesses ranging from heart disease to Alzheimer's. In addition to curing diseases, gene therapy (gene editing) could be used to stop inherited disease in its tracks to save endangered species, and more so, to resurrect extinct species. All such gene therapy techniques are clinical laboratory-based. Alternatively, one can try out the possibilities of developing some of them for genome editing using computational tools and concepts. Genome is a string of nucleotides, and its characteristic sequence is a sequence of A, T, G, and C. Gene editing, in the computer science point of view, is a pattern searching and substituting process, meaning, genome editing is essentially a string processing operation. A genome is a subset of a free monoid A* of an alphabet A which consists of the primitive symbols A, T, G and C. So, gene editing is viewed as a map ? that connects A* to A*.It is in this context, this paper introduces a novel concept of genome editing cellular automata evolutions of adjoint strings of a genome. Section 2 of this paper describes the fundamental notions of adjoints of a genome and their evolution using one-dimensional cellular automata rules defined by linear Boolean functions. Section 3 presents the technique of editing genome sequences using the cellular automata generations of adjoint sequences. Section 4 illustrates the concept with the help of a case study. II. Adjoints of A Genome Sequence and their Evolutions using one-Dimensional three Neighborhood Cellular Automata Adjoint of a particular nucleotide in a genome sequence is the binary sequence obtained by substituting the particular nucleotides in the genome sequence by 1's and the others by 0's. For example, let us consider a sample sequence of BrucellaSuis 1330 for a case study. The actual length of the genome sequence of BrucellaSuis 1330 is 5806. A cellular automaton is an idealized parallel processing system consisting of an array of numbers (1-D, 2-D and more) realized using updating rules based on certain Consider an ith cell in the array. This cell has a neighbor i-1 on its left and another i+1 on its right. All three put together is called a three neighborhood. One can assign a site (cell) variable ?i-1, ?i, and ?i+1 to the three neighborhood cells. At a particular instant of time, these variables take on numerical values, say either a 0 or a 1. In such a case, the variables are denoted as ?ti-1, ?ti, and ?ti+1. The value of the ith cell at the next instant of time is evaluated using an updating rule that involves the present values of the ith, (i-1)th and (i+1)th cells. This updating rule is basically a linear Boolean function of three variables. One can construct 256 linear Boolean functions, as updating rules of one-dimensional threeneighborhood binary-valued cellular automata. Each rule defines an automaton by itself. So, one dimensional binary-valued three-neighborhood cellular automata (123CA) rules could be used to model adjoints of a genome sequence. The first twenty linear Boolean functions of cellular automata 123CA are listed below with their decimal equivalents. # Linear Boolean Function Decimal Equivalent 0 0 (?? ? ???1 ?? ? ?? ?? ? ??+1 ) 1 (?? ? ???1 ?? ? ?? ?? ??+1 ) 2 (?? ? ???1 ?? ? ?? ) 3 (?? ? ???1 ?? ?? ?? ? ??+1 ) 4 (?? ? ???1 ?? ? ??+1 ) 5 (?? ? ???1 ?? ?? ?? ? ??+1 )+(?? ? ???1 ?? ? ?? ?? ??+1 ) 6 (?? ? ???1 ?? ? ??+1 ) + (?? ? ???1 ?? ? ?? ) 7 (?? ? ???1 ?? ?? ?? ??+1 ) 8 (?? ? ???1 ?? ? ?? ?? ? ??+1 ) + (?? ? ???1 ?? ?? ?? ??+1 ) 9 (?? ? ???1 ?? ??+1 ) 10 (?? ? ???1 ?? ? ?? ) + (?? ? ???1 ?? ??+1 ) 11 (?? ? ???1 ?? ?? ) 12 (?? ? ???1 ?? ? ??+1 ) + (?? ? ???1 ?? ?? ) 13 (?? ? ???1 ?? ?? ) + (?? ? ???1 ?? ??+1 ) 14 (?? ? ???1 ) 15 (?? ???1 ?? ? ?? ?? ? ??+1 ) 16 (?? ? ?? ?? ? ??+1 ) 17 (?? ???1 ?? ? ?? ?? ? ??+1 ) + (?? ? ???1 ?? ? ?? ?? ??+1 ) 18 (?? ? ?? ?? ? ??+1 ) + (?? ? ???1 ?? ? ?? ) 19 (?? ???1 ?? ? ?? ?? ? ??+1 ) + (?? ? ???1 ?? ?? ?? ? ??+1 )20 For the case study rule number 90 is applied to the adjoints of BrucellaSuis 1330 genome sequence and 500 evolutions generated. Rule 90 is shown below. (? ???1 ? ? ??+1 ) + (? ? ???1 ? ??+1 )90 Since the image of the 500 evolutions of BrucellaSuis 1330 is quite large, a small portion of the images are presented in this paper. Fig. 1 shows evolutions of the adjoints of A(n) and T(n). As outlined earlier, a genome sequence is a subset of a free monoid A* of an alphabet A which consists of the primitive symbols A, T, G and C and gene editing is a map ? that connects A * to A *. The map ? is a rule that transforms a sequence into another desired sequence. One can use four types of symbol to symbol substitution formulas given below for genome editing. # Symbol to be substituted Substitution symbol Formula Number A?T?G?C ? A 1 A?T?G?C ? T 2 A?T?G?C ? G 3 A?T?G?C ? C 4 Where the symbol ? denotes the relation of logical OR. Application of formula #1 to any genome sequence is called A-latch. Similarly one can think of Tlatch, G-latch and C-latch. A nucleotide latch would give rise to conversion of any genome sequence into a sequence consisting of that particular nucleotide in one step. This is called 'Nucleotide Saturation'. Thus one can have A-saturation, T-saturation,G-saturation,and Csaturation. To be precise, A-latching of any genome sequence transforms it into A-saturated sequence and it holds for other three types of latching also. The question that arises here is that whether it is possible to edit a genome sequence using nucleotide latching techniques at preferred locations in the sequence. One such possibility is latching of a genome sequence using cellular automata evolutions of adjoints. Once a genome sequence is latched, the resulting edited sequence is called its elementary transformation. Two types of latching are discussed here (i) latching of elementary transformations of a genome sequence with various cellular automata evolutions of adjoints and (ii) latching of original genome sequence with various cellular automata evolutions of adjoints. Fig. 3 portraits the first approach of latching of elementary transformations of a genome sequence with various cellular automata evolutions of adjoints. # IV. Case Study The characteristic sequence of BrucellaSuis 1330 genome sequence is used here for the case study. The length of this sequence is 5806. Rule number 90 is used here for the study. The linear Boolean function corresponding to this rule is (? ???1 ? ? ??+1 ) + (? ? ???1 ? ??+1 ) Approach #1 Editing of elementary transformations of a genome sequence with various cellular automata evolutions of adjoints Figs. 5 to 8 portrait the first approach of latching of elementary transformations of a genome sequence with various cellular automata evolutions of adjoints. Fig. 5 shows the result of A-latching the genome sequence, that is, Adenine based genome editing. Since the image is quite large, a small portion of it is shown here. It was observed that the A-saturation of the genome sequence occurred while editing the previous elementary transformation of the genome sequence using the 32nd evolution of the A(n). Elementary transformations of the genome sequence are edited using the rule number 90 based cellular automaton evolution of A (n). A-saturation occurred when the 32 nd evolution is used for editing. G-saturation occurred when the 15 th evolution is used for editing. Approach #2 Editing of a genome sequence with various cellular automata evolutions of adjoints Figs. 9 to 14 portrait the second approach of latching of a given genome sequence with rule number 90 based cellular automata evolutions of adjoints. Figs. 9 shows Adenine based genome editing on original sequence both in Text Form and in Image Form. The Image Form is obtained using a color coding scheme which paints a red color for Adenine, green color for Thymine, blue color for Guanine and yellow color for Cytosine. Fig. 10 shows the predominant horizontal lines and vertical lines of the image of the edited genome separately. An image processing tool of line detector is used for this purpose. The following observations are made from the case study. 1. Editing a genome is a map or a function that transforms a genome into a desired genome sequence 2. Two approaches could be undertaken for genome editing (i) Editing of elementary transformations of a genome sequence with various cellular automata evolutions of adjoints and(ii) Editing of the given genome sequence with various cellular automata evolutions of adjoints 3. Using approach 1, one would end up with saturation in short steps. Using approach 2, one would be able to generate edited versions that exhibit periodicities and generic evolutions. 5. The result of each operation exhibits unique behavior in periodicity and generic form. 6. Interpretation of these properties could be made in a better way only by an expert in genetic Engineering 2![Fig. 1: Evolutions of A(n) and T(n) Fig. 2 shows evolution of the adjoints of G(n) and C(n).](image-2.png "Fig. 2 :") 3![Fig. 3: Editing of elementary transformations of a genome sequence Fig. 4 portraits the second approach of latching of original genome sequence with various cellular automata evolutions of adjoints.](image-3.png "Fig. 3 :") 4![Fig. 4: Editing of original genome sequence](image-4.png "Fig. 4 :") 5![Fig. 5: Adenine based genome editing (A-saturation at 32 nd evolution) Fig. 6 shows the result of T-latching the genome sequence, that is, Thymine based genome editing. Since the image is quite large, a small portion of it is shown here. It was observed that the T-saturation of the genome sequence occurred while editing the previous elementary transformation of the genome sequence using the 40th evolution of the A(n).Elementary transformations of the genome sequence are edited using the rule number 90 based cellular automaton evolution of T(n). T-saturation occurred when the 40 th evolution is used for editing.](image-5.png "Fig. 5 :") 6![Fig. 6: Thymine based genome editing (T-saturation at 40 th evolution)Fig.7shows the result of G-latching the genome sequence, that is, Guanine based genome editing. Since the image is quite large, a small portion of it is shown here. It was observed that the G-saturation of the genome sequence occurred while editing the previous elementary transformation of the genome sequence using the 15th evolution of the G(n).](image-6.png "Fig. 6 :") ![Computer Science and Technology Volume XX Issue I Version I Year Elementary transformations of the genome sequence are edited using the rule number 90 based cellular automaton evolution of GT(n).](image-7.png "") 7![Fig. 7: Guanine based genome editing (G-saturation at 15 th evolution)Fig.8shows the result of C-latching the genome sequence, that is, Cytosine based genome editing. Since the image is quite large, a small portion of it is shown here. It was observed that the C-saturation of the genome sequence occurred while editing the previous elementary transformation of the genome sequence using the 21stevolution of the C(n).](image-8.png "Fig. 7 :") 8![Fig. 8: Cytosine based genome editing (C-saturation at 21 st evolution)](image-9.png "Fig. 8 :") 91011![Fig. 9: Edited genome using evolutions of A(n)](image-10.png "Fig. 9 :Fig. 10 :Fig. 11 :") 121314151617181920212223242526![Fig.12: Horizontal and Vertical lines of T-latch detected](image-11.png "Fig. 12 :Fig. 13 :Fig. 14 :Fig. 15 :Fig. 16 :Fig. 17 :Fig. 18 :Fig. 19 :Fig. 20 :Fig. 21 :Fig. 22 :Fig. 23 :Fig. 24 :Fig. 25 :Fig. 26 :") 272728![Fig.27shows the result of comparing T-latch and C-latch of BrucellaSuis 1330 genome with the min operator along with its horizontal and vertical lines. Fig.28shows the result of comparing G-latch and C-latch of BrucellaSuis 1330 genome with the min operator along with its horizontal and vertical lines.](image-12.png "Fig. 27 Fig. 27 :Fig. 28 :") 4![Relational operations like 'max' and 'min' could be carried out on various latches. One can have 11 such max operations like (i) A-latch and T-latch, (ii) A-latch and G-latch, (iii) A-latch and C-latch, (iv) Tlatch and G-latch, (v) T-latch and C-latch, (vi) Glatch and C-latch, (vii) A-latch, T-latch and G-latch (viii) A-latch, T-latch and C-latch(ix) A-latch, G-latch and C-latch, (x)T-latch, G-latch and C-latch, (xi) Alatch,T-latch, G-latch and C-latch. In the same manner, one can have 11 min operations.](image-13.png "4 .") © 2020 Global Journals ## Acknowledgements The authors express their profound gratitude to the management of Pentagram Research Centre Private Limited, Hyderabad for their boundless support in carrying out research in their premises. The authors further put on record the invaluable guidance of Prof. Dr E G Rajan, President of Pentagram Research and data scientists Mr. Rahul Sharma, Mr. Srikanth Maddikunta and Mr. Shubham Karande. * Worlds Record Breaking Plant: Deletes its Noncoding "Junk" DNA". Design & Trend May 12. 2013 References Références Referencias 1. Retrieved 2013-06-04 * ENCODE Project Writes Eulogy for Junk DNA EPennisi 10.1126/science.337.6099.1159 22955811 Science 337 6099 6 September 2012 * An integrated encyclopedia of DNA elements in the human genome 10.1038/nature11247.PMC3439153 22955616 Bibcode: 2012Natur.489...57T 2012 489 * Non-coding RNAs and Epigenetic Regulation of Gene Expression: Drivers of Natural Selection FabricoCosta 2012. 1904455948 Caister Academic Press Morris, Kevin V 7 Non-coding RNAs, Epigenomics, and Complexity in Human Cells * Junk DNA: A Journey Through the Dark Matter of the Genome NessaCarey 2015 Columbia University Press ISBN 9780231170840 * Scientists attacked over claim that 'junk DNA' is vital to life RobinMckie 24 February 2013 The Observer * The C-value paradox, junk DNA, and ENCODE SeanEddy CurrBiol 22 21 2012 * Is junk DNA bunk? A critique of ENCODE WDoolittle Ford 10.1073/pnas.1221376110.PMC3619371 23479647 Bibcode:2013PNAS..110.5294D 2013 110 * The Case for Junk DNA AlexanderFPalazzo TGregory Ryan 10.1371/journal.pgen.1004351.ISSN1553-7404 PLoS 10 5 e1004351 2014 * On the immortality of television sets: "function DanGraur YichenZheng NicholasPrice RicardoB RAzevedo1 RebeccaAZufall Eranelhaik 2013 in the human genome according to the * Computer Science and Technology Volume XX Issue I Version I Year evolution-free gospel of ENCODE" (PDF) 10.1093/gbe/evt028.3622293 23431001 Genome Biology 5 3 * CPPonting * What fraction of the human genome is functional? RCHardison 10.1101/gr.116814.110.PMC3205562 21875934 Genome 21 2011 * Defining functional DNA elements in the human genome MKellis 10.1073/pnas.1318948111.PMC4035993 24753594 Bibcode: 2014PNAS..111.6131K 2014 111 * 8.2% of the Human Genome Is Constrained: Variation in Rates of Turnover across Functional Element Classes in the Human Lineage ChrisMRands StephenMeader ChrisPPonting Gertonlunter 10.1371/journal.pgen.1004525.PMC4109858 25057982 PLoS Genet 10 7 e1004525 2014 * The extent of functionality in the human genome JSMattick MEDinger 10.1186/1877-6566-7-2 The HUGO Journal 7 1 2 2013 * Non-Coding RNAs and Epigenetic Regulation of Gene Expression: Drivers of Natural Selection KevinMorris 2012. 1904455948 Caister Academic Press Norfolk, UK * GElgar TVavouri * Tuning in to the signals: noncoding sequence conservation in vertebrate genomes Vavouri 10.1016/j.tig.2008.04.005 18514361 Trends Genet 24 7 July 2008 * TRGregory PDHebert * The modulation of DNA content: proximate causes and ultimate consequences Hebert 10.1101/gr.9.4.317(inactive2015-02-01 10207154 Genome Res 9 4 April 1999 * Hypervariable minisatellite DNA is a hotspot for homologous recombination in human cells WPWahls 10.1016/0092-8674(90)90719-UPMID2295091 Cell 60 1 1990 * Plant biology: Coding in non-coding RNAs PeterMWaterhouse RogerPHellens 10.1038/nature14378 Nature 520 7545 25 March 2015 * MLi CMarin-Muller UBharadwaj KHChow QYao C;Chen ;Marin-Muller ;Bharadwaj ;Chow Yao * MicroRNAs: Control and Loss of Control in Human Physiology and Disease Chen 10.1007/s00268-008-9836-x.PMC2933043 19030926 World J 33 4 April 2009 * Genomic Views of Distant-Acting Enhancers AVisel EMRubin LAPennacchio Nature 461 7261 September 2009 * 10.1038/nature08451.PMC2923221.PMID19741700 * HNielsen SDJohansen * Group I introns: Moving in new directions Johansen 10.4161/rna.6.4.9334 19667762 RNA 6 4 2009 * Pseudogenes in the ENCODE regions: Consensus annotation, analysis of transcription, and evolution DZheng AFrankish RBaertsch June 2007 * 17568002 PMC1891343 Genome Res 17 6 * Dollo's law and the death and resurrection of genes CRMarshall ECRaff RARaff ;Raff Raff 7991619 PMC45421 Bibcode:1994PNAS...9112283M December 1994 91 * Pseudogenes YTutar 10.1155/2012/424526.PMC3352212 22611337 Comp Funct Genomics 424526 2012. 2012 * DAPetrov DLHartl * Pseudogene evolution and natural selection for a compact genome Hartl 10833048 J. Hered 91 3 2000 * SLPonicsan JFKugel Ja;Goodrich Kugel * Genomic gems: SINE RNAs regulate mRNA production Goodrich 10.1016/j.gde.2010.01.004.PMC2859989 20176473 Current Opinion in Genetics & Development 20 2 February 2010 * JHäsler TSamuelsson K;Strub Samuelsson * Useful 'junk': Alu RNAs in the human transcriptome Strub 17514354 Cell. Mol. Life Sci 64 14 July 2007 * RDWalters JFKugel Ja;Goodrich Kugel * InvAluable junk: the cellular impact and function of Alu and B2 RNAs Goodrich 10.1002/iub.227.PMC4049031 19621349 IUBMB Life 61 8 Aug 2009 * Human endogenous retroviruses: transposable elements with potential? PNNelson PHooley DRoden HDavariejtehadi PRylance PWarren JMartin PGMurray 10.1111/j.1365-2249.2004.02592.x.PMC1809191.PMID15373898.31 11237011 International Human Genome Sequencing Consortium Oct 2004. February 2001 138 Clin. Bibcode: 2001Natur.409..860L. * Doubling genome size without polyploidization: dynamics of retrotransposition-driven genomic expansions in Oryzaaustraliensis BPiegu RGuyot NPicault ARoulin ASanyal ASaniyal HKim KCollura Oct 2006 a wild relative of rice * 10.1101/gr.5290206.1581435 16963705 Genome Res 16 10 * Differential lineage-specific amplification of transposable elements is responsible for genome size variation in Gossypium JSHawkins HKim JDNason RAWing JFWendel 10.1101/gr.5282906.1581434 16954538 Genome Res 16 10 Oct 2006 * So Much "junk SusumuOhno DNA in Our Genome. Gordon and Breach H. H. Smith 1972. 2013-05-15 * Selfish genes, the phenotype paradigm and genome evolution WFDoolittle C;Sapienza Sapienza Nature 284 5757 1980 * 10.1038/284601a0 6245369 * Another source is genome duplication followed by a loss of function due to redundancy * Selfish DNA: the ultimate parasite LEOrgel Fh;Crick Crick 10.1038/284604a0 7366731 Nature 284 April 1980 Bibcode:1980Natur.284..604O. * AKhajavinia WMakalowski * The term "junk DNA" repelled mainstream researchers from studying non coding genetic material for many years Makalowski 17503549 Scientific 296 5 104 May 2007 What is "junk" DNA, and what is it worth? * Genetics: Junk DNA as an evolutionary force Christian;Biémont CVieira 17024082 Bibcode:2006Natur.443..521B 2006 443 * Noncoding RNA: what is functional and what is junk? AlexanderFPalazzo ElizaSLee 10.3389/fgene.2015.00002 25674102 Frontiers in 6 2 2015 * Functional evolution of noncoding DNA MZLudwig 10.1016/S0959-437X(02)00355-6 12433575 Current Opinion in Genetics & Development 12 6 December 2002 * JCobb CBüsst SPetrou SHarrap J;Ellis Büsst * ;Petrou Harrap * Searching for functional genetic variants in non-coding DNA Ellis 18307723 Clin. Exp. Pharmacol. Physiol 35 4 April 2008 * Integrative annotation of variants from 1092 humans: application to cancer genomics ;Khurana 10.1126/science.1235587.PMC3947637 24092746 Science 342 6154 April 2013 * Yi-FanLu * IFNL3 mRNA structure is remodeled by a functional non-coding polymorphism associated with hepatitis C virus clearance DavidMMauger DavidBGoldstein ThomasJUrban KevinMWeeks SheltonSBradrick 10.1038/srep16037 26531896 Scientific Reports 5 16037 4 November 2015 * ThomasGGrünewald * Virginie;Bernard Pascale;Gilardi-Hebenstreit Virginie;Raynal Surdez ;Didier Marie-MingAynaud * OlivierMirabeau * Chimeric EWSR1-FLI1 regulates the Ewing sarcoma susceptibility gene EGR2 via a GGAA microsatellite Florencia;Cidre-Aranaz FranckTirode 10.1038/ng.3363 26214589 PMC4591073 Nature Genetics 47 9 * The most frequent short sequences in noncoding DNA JASubirana X;Messeguer Messeguer 10.1093/nar/gkp1094.PMC2831315 PMID Nucleic Acids Res 38 4 March 2010. 19966278 * SEAhnert * How much noncoding DNA do eukaryotes require?" (PDF) TM AFink doi:10 J. Theor. Biol 252 4 2008 * Widespread purifying selection on RNA structure in mammals MASmith 23847102 PMC3783177 Nucleic Acids Research 41 17 June 2013 * The place and function of noncoding DNA in the evolution of variability VDileep doi:10.5779/ hypothesis. v7i1.146 Hypothesis 7 1 e7 2009 * Regulating Evolution SeanBCarroll 10.1038/scientificamerican0508-60 18444326 Scientific American 298 5 May 2008 * Transcriptional silencing of long noncoding RNA GNG12-AS1 uncouples its transcriptional and product-related functions LStojic nature.com. Nature. Retrieved 21 Feb 2016 * Junk DNA gets credit for making us who we are EwenCallaway New Scientist Edward E. Max, M.D., Ph.D. 57 56 March 2010 Ayala FJ * Pseudogenes: are they "junk" or functional DNA? Ayala 14616058 Annu. Rev. Genet 37 2003 * C.-KPeng SVBuldyrev ALGoldberger SHavlin FSciortino MSimons HEStanley * Long-range correlations in nucleotide sequences SVBuldyrev Goldberger S;Al; Havlin F;Sciortino M;Simons Stanley He 10.1038/356168a0 1301010 Bibcode:1992Natur.356..168P 1992 356 * WLi ,KKaneko * Long-Range Correlation and Partial 1/f alpha Spectrum in a Non-Coding DNA Sequence" (PDF) KKaneko Bibcode:1992EL.....17..655L 10.1209/02955075/17/7/014 Europhys. Lett 17 7 1992 * Long-range correlations properties of coding and noncoding DNA sequences: GenBank analysis SVBuldyrev ALGoldberger SHavlin RNMantegna MMatsa C.-KPeng MSimons HEStanley; Goldberger AHavlin SMantegna RMatsa MPeng C.-KSimons MStanley H 10.1103/PhysRevE.51.5084 Phys. Rev. E 51 5 1995. 1995PhRvE