Heuristics for similarity searching of chemical graphs using a maximum common edge subgraph algorithm

Recently a method (RASCAL) for determining graph similarity using a maximum common edge subgraph algorithm has been proposed which has proven to be very efficient when used to calculate the relative similarity of chemical structures represented as graphs. This paper describes heuristics which simplify a RASCAL similarity calculation by taking advantage of certain properties specific to chemical graph representations of molecular structure. These heuristics are shown experimentally to increase the efficiency of the algorithm, especially at more distant values of chemical graph similarity.     

Graph edge colorings and their chemical applications

A general method is outlined to enumerate the edge-colorings of graphs under group action. The symmetry group of the graph acting on the vertices induces permutation of the edges. The edge-colorings are enumerated using the edge-permutation group. A number of chemical applications especially to multiple quantum NMR spectroscopy, statistical mechanics, enumeration of unsaturated isomers, etc. are considered.Alfred P. Sloan fellow; Camille and Henry Dreyfus teacher-scholar     

Method for construction of characteristic polynomials via graph linearization

A new method for construction of characteristic polynomials (CP) of complicated graphs having arbitrary edge and vertex weights has been developed. The method first converts the graph into isospectral linear chains with weighted vertices and edges and then builds up the CP coefficients recursively. Two types of graphs have been used for illustration, viz., (i) graphs that can be linearized by symmetry factorization and (ii) graphs without symmetry which are to be linearized by an algorithm involving walks of unit length. Both types have been illustrated, of which type (i) includes the Schlegel of fullerene fragment C₂₀ and another large graph with many fused rings. © 1997 John Wiley & Sons, Inc. Int J Quant Chem 65 : 199–204, 1997     

Graph theoretical procedure for obtaining analytical expressions of eigenspectra of linear chains and cycles with alternant vertex weights and same edge weight: Application to some complicated graphs

A graph theoretical procedure for obtaining eigenvalues of linear chains and cycles having alternant vertex weights (h₁, h₂, h₁, h₂, h₁, h₂, …) and the same edge weight (k) have been developed. The eigenvalues of some complicated graphs, such as graphs of linear polyacenes, methylene‐substituted linear polyacenes and cylindrical polyacene strips, stack graphs, and reciprocal graphs have been shown to be generated in closed analytical forms by this procedure. Many such graphs represent chemically important molecules or radicals. © 2005 Wiley Periodicals, Inc. Int J Quantum Chem, 2005     

Computer generation of spectra of graphs: Applications to C60 clusters and other systems

A computer code based on the Givens–Householder matrix diagonalization method is used to calculate the spectra of graphs containing a large number of vertices. The code is most general in that it can handle graphs containing 200 or more vertices. Further, the code can be used to generate the spectra of weighted graphs. The program requires as input only the neighborhood table of the graph. The spectra of many graphs are generated for the first time in less than a few minutes of computer time. Applications to a number of chemical systems including two forms (foot and hand) of the recently synthesized C₆₀ cluster and the effect of bond alternation on these systems are discussed. In addition, the spectra of square and honeycomb lattices and the characteristic polynomials of the foot and hand forms of the C₆₀ cluster are obtained.     

On the centrality of vertices of molecular graphs

For acyclic systems the center of a graph has been known to be either a single vertex of two adjacent vertices, that is, an edge. It has not been quite clear how to extend the concept of graph center to polycyclic systems. Several approaches to the graph center of molecular graphs of polycyclic graphs have been proposed in the literature. In most cases alternative approaches, however, while being apparently equally plausible, gave the same results for many molecules, but occasionally they differ in their characterization of molecular center. In order to reduce the number of vertices that would qualify as forming the center of the graph, a hierarchy of rules have been considered in the search for graph centers. We reconsidered the problem of “the center of a graph” by using a novel concept of graph theory, the vertex “weights,” defined by counting the number of pairs of vertices at the same distance from the vertex considered. This approach gives often the same results for graph centers of acyclic graphs as the standard definition of graph center based on vertex eccentricities. However, in some cases when two nonequivalent vertices have been found as graph center, the novel approach can discriminate between the two. The same approach applies to cyclic graphs without additional rules to locate the vertex or vertices forming the center of polycyclic graphs, vertices referred to as central vertices of a graph. In addition, the novel vertex “weights,” in the case of acyclic, cyclic, and polycyclic graphs can be interpreted as vertex centralities, a measure for how close or distant vertices are from the center or central vertices of the graph. Besides illustrating the centralities of a number of smaller polycyclic graphs, we also report on several acyclic graphs showing the same centrality values of their vertices. © 2013 Wiley Periodicals, Inc.     

Computational algorithms for matching polynomials of graphs from the characteristic polynomials of edge-weighted graphs

Computational algorithms are described which provide for constructing the set of associated edge-weighted directed graphs such that the average of the characteristic polynomials of the edge-weighted graphs gives the matching polynomial of the parent graph. The weights were chosen to be unities or purely imaginary numbers so that the adjacency matrix is hermitian. The computer code developed earlier by one of the authors (K.B.) is generalized for complex hermitian matrices. Applications to bridged and spirographs, some lattices and all polycyclic graphs containing up to four cycles are considered.     

On the geometric–arithmetic index by decompositions-CMMSE

The concept of geometric–arithmetic index was introduced in the chemical graph theory recently, but it has shown to be useful. There are many papers studying different kinds of indices (as Wiener, hyper–Wiener, detour, hyper–detour, Szeged, edge–Szeged, PI, vertex–PI and eccentric connectivity indices) under particular cases of decompositions. The main aim of this paper is to show that the computation of the geometric-arithmetic index of a graph G is essentially reduced to the computation of the geometric-arithmetic indices of the so-called primary subgraphs obtained by a general decomposition of G. Furthermore, using these results, we obtain formulas for the geometric-arithmetic indices of bridge graphs and other classes of graphs, like bouquet of graphs and circle graphs. These results are applied to the computation of the geometric-arithmetic index of Spiro chain of hexagons, polyphenylenes and polyethene.     

Pairing theorem of graph eigenvalues: Its new proof and a generalization

Each undirected graph has its own adjacency matrix, which is real and symmetric. The negative of the adjacency matrix, also real and symmetric, is a well-defined mathematically elementary concept. By this negative adjacency matrix, the negative of a graph can be defined. Then an orthogonal transformation can be readily found that transforms a negative of an alternant graph to that alternant graph: (?G) → G. Since the procedure does not involve the edge weights, the pairing theorem holds true for all edge-weighted alternant graphs, including the usual “standard” graphs.     

Computer generation of distance polynomials of graphs

A computer program is developed to compute distance polynomials of graphs containing up to 200 vertices. The code also computes the eigenvalues and the eigenvectors of the distance matrix. It requires as input only the neighborhood information from which the program constructs the distance matrix. The eigenvalues and eigenvectors are computed using the Givens-Householder method while the characteristic polynomials of the distance matrix are constructed using the codes developed by the author before. The newly developed codes are tested out on many graphs containing large numbers of vertices. It is shown that some cyclic isospectral graphs are differentiated by their distance polynomials although distance polynomials themselves are in general not unique structural invariants.     

Structural data representation and search for structural analogs in molecular spectroscopy databases

A method for encoding structural formulas of organic compounds using two varieties of the tree code is suggested. The tree code is a sequence of symbols, each corresponding to an edge of a molecular graph traced around in width or in depth. The deep tree code provides compact storage of structural formulas and rapid substructure search. The wide tree code is a basis for a structure collection classifier providing quick access to structurally related compounds. Three techniques for selecting structural analogs of the compound using the classifier are proposed. Databases on mass and¹³C NMR spectra serve as illustrations. Novosibirsk Institute of Organic Chemistry, Siberian Branch, Russian Academy of Sciences. Translated fromZhurnal Strukturnoi Khimii, Vol. 37, No. 6, pp. 1129–1139, November–December, 1996. Translated by L. Smolina     

On discerning symmetry properties of graphs

A procedure is outlined which allows the symmetry properties of graphs to be systematically and rigorously investigated. It is based on a search for all the automorphisms of a graph and this is accompanied by suitably applying the procedure for recognizing identical graphs. It consists in finding all the distinctive labeling of the vertices of graph associated with the smallest binary code derived by a particular interpretation of the associated adjacency matrix. No prior cognizance of symmetry operations is required which is in contrast to the usual discussions of the symmetry properties of molecules which are based on the knowledge of pertinent symmetry operations. This is important since in graphs it is neither apparent nor generally possible to simply enlist those permutations of labels which leave the connectivity invariant (i.e., do not alter the form of the adjacency matrix). The procedure is applied to the Petersen graphs and the Desargues-Levi graph, both associated with isomerizations of trigonal bipyramidal complex and other chemical transformations. It is shown that these graphs of high symmetry belong to symmetry groups of order 120 and 240 respectively. The approach can also provide a basis for the development of the symmetry properties of non-rigid molecules in which connectivity is preserved.     

Computer enumeration of walks on directed graphs

A vectorized computer code is developed for the enumeration of walks through the matrix power method for directed graphs. Application of this code to several graphs is considered. It is shown that the coefficients in the generating functions for signed graphs are much smaller in magnitude. It is shown that self-avoiding walks on some graphs can be enumerated as a linear combination of walk GFs of directed paths and rooted-directed paths.     

Normal Components, Kekulé Patterns, and Clar Patterns in Plane Bipartite Graphs

As a general case of molecular graphs of polycyclic alternant hydrocarbons, we consider a plane bipartite graph G with a Kekulé pattern (perfect matching). An edge of G is called nonfixed if it belongs to some, but not all, perfect matchings of G. Several criteria in terms of resonant cells for determining whether G is elementary (i.e., without fixed edges) are reviewed. By applying perfect matching theory developed in plane bipartite graphs, in a unified and simpler way we study the decomposition of plane bipartite graphs with fixed edges into normal components, which is shown useful for resonance theory, in particular, cell and sextet polynomials. Further correspondence between the Kekulé patterns and Clar (resonant) patterns are revealed.     

Scaffold hopping using clique detection applied to reduced graphs

Similarity-based methods for virtual screening are widely used. However, conventional searching using 2D chemical fingerprints or 2D graphs may retrieve only compounds which are structurally very similar to the original target molecule. Of particular current interest then is scaffold hopping, that is, the ability to identify molecules that belong to different chemical series but which could form the same interactions with a receptor. Reduced graphs provide summary representations of chemical structures and, therefore, offer the potential to retrieve compounds that are similar in terms of their gross features rather than at the atom-bond level. Using only a fingerprint representation of such graphs, we have previously shown that actives retrieved were more diverse than those found using Daylight fingerprints. Maximum common substructures give an intuitively reasonable view of the similarity between two molecules. However, their calculation using graph-matching techniques is too time-consuming for use in practical similarity searching in larger data sets. In this work, we exploit the low cardinality of the reduced graph in graph-based similarity searching. We reinterpret the reduced graph as a fully connected graph using the bond-distance information of the original graph. We describe searches, using both the maximum common induced subgraph and maximum common edge subgraph formulations, on the fully connected reduced graphs and compare the results with those obtained using both conventional chemical and reduced graph fingerprints. We show that graph matching using fully connected reduced graphs is an effective retrieval method and that the actives retrieved are likely to be topologically different from those retrieved using conventional 2D methods.     

Computer generation of characteristic polynomials of edge-weighted graphs,heterographs, and directed graphs

The computer code developed previously (K. Balasubramanian, J. Computational Chem., 5 , 387 (1984)) for the characteristic polynomials of ordinary (nonweighted) graphs is extended in this investigation to edge-weighted graphs, heterographs (vertex-weighted), graphs with loops, directed graphs, and signed graphs. This extension leads to a number of important applications of this code to several areas such as chemical kinetics, statistical mechanics, quantum chemistry of polymers, and unsaturated systems containing heteroatoms which include bond alternation. The characteristic polynomials of several edgeweighted graphs which may represent conjugated systems with bond alternations, heterographs (molecules with heteroatoms), directed graphs (chemical reaction network), and signed graphs and lattices are obtained for the first time.     

On reciprocal reverse Wiener index

We report some properties, especially bounds for the reciprocal reverse Wiener index of a connected (molecular) graph. We find that the reciprocal reverse Wiener index possesses the minimum values for the complete graph in the class of n-vertex connected graphs and for the star in the class of n-vertex trees, and the maximum values for the complete graph with one edge deleted in the class of n-vertex connected graphs and for the tree obtained by attaching a pendant vertex to a pendant vertex of the star on n − 1 vertices in the class of n-vertex trees. These results are compared with those obtained for the ordinary Wiener index.