Predicting properties of molecules using graph invariants

Topological indices (TIs) have been used to study structure-activity relationships (SAR) with respect to the physical, chemical, and biological properties of congeneric sets of molecules. Since there are many TIs and many are correlated, it is important that we identify redundancies and extract useful information from TIs into a smaller number of parameters. Moreover, it is important to determine if TIs, or parameters derived from TIs, can be used for global SAR models of diverse sets of chemicals. We calculated seventy-one TIs for three groups of molecules of increasing complexity and diversity: (a) 74 alkanes, (b) 29 alkylbenzenes, and (c) 37 polycyclic aromatic hydrocarbons (PAHs). Principal components analysis (PCA) revealed that a few principal components (PCs) could extract most of the information encoded by the seventy-one TIs. The structural basis of the first few PCs could be derived from their pattern of correlation with individual TIs. For the three sets of molecules, viz. alkanes, alkylbenzenes and PAHs, PCs were able to predict the boiling points reasonably well. Also, for the combined set of 140 chemicals consisting of the alkanes, alkylbenzenes and PAHs, the derived PCs were not as effective in predicting properties as in the case of individual classes of compounds.     

Optimal neighbor selection in molecular similarity: comparison of arbitrary versus tailored prediction spaces

Three classes of arbitrary quantitative molecular similarity analysis (QMSA) methods have been computed using atom pairs (APs), topological indices (TIs), and principal components (PCs) derived from topological indices. Tailored QMSA models have been developed from TIs selected through ridge regression. K-nearest neighbor (kNN) based estimation has been applied to all of the methods to estimate normal vapor pressure (p(vap)) and water solubility (sol) for a set of 194 chemicals. Results show that the tailored QMSA methods are superior to arbitrary similarity methods in estimating both of these properties for the given set of chemicals.     

Quantitative structure-activity relationship studies of boron-containing dipeptide proteasome inhibitors using calculated mathematical descriptors

Topological indices (TIs) and atom pairs (APs) were used to develop quantitative structure-activity relationship (QSAR) models of a set of 58 dipeptide boronic acids which are potent inhibitors of proteasome and have found applications in the treatment of various types of cancers. Of the three linear regression methods used for QSAR development, viz., principal components regression (PCR), partial least square (PLS), and ridge regression (RR), the last method gave the most satisfactory models whereas the remaining two methods yielded poor models. RR results obtained in this paper using TIs and APs are comparable to the CoMFA and CoMSIA results reported in the literature with the same set of compounds.     

Three dissimilarity measures to contrast dendrograms

We discussed three dissimilarity measures between dendrograms defined over the same set, they are triples, partition, and cluster indices. All of them decompose the dendrograms into subsets. In the case of triples and partition indices, these subsets correspond to binary partitions containing some clusters, while in the cluster index, a novel dissimilarity method introduced in this paper, the subsets are exclusively clusters. In chemical applications, the dendrograms gather clusters that contain similarity information of the data set under study. Thereby, the cluster index is the most suitable dissimilarity measure between dendrograms resulting from chemical investigation. An application example of the three measures is shown to remark upon the advantages of the cluster index over the other two methods in similarity studies. Finally, the cluster index is used to measure the differences between five dendrograms obtained when applying five common hierarchical clustering algorithms on a database of 1000 molecules.     

Optimal characterization of structure for prediction of properties

Different topological and physicochemical parameters have been used to predict hydrophobicity (logP, octanol-water) of chemicals. We calculated a hydrogen bonding parameter (HB₁) and a large number of molecular connectivity and complexity indices for a diverse set of 382 molecules. It is known from earlier studies that topological indices (TIs) predict properties of congeneric sets reasonably well. Since HB₁ is an approximate quantifier of hydrogen bonding and has integral values, we used HB₁ to classify the diverse set into strongly and weakly hydrogen bonding subsets. In an attempt to examine the utility of Us in predicting properties of relatively similar groups of molecules, we carried out a correlation of logP with TIs for a subset (n = 139) of the original diverse set (n = 382) with a weak hydrogen bonding ability (HB₁ = 0). Results show that TIs give a better predictive model for the more homogeneous subset as compared to the diverse set of molecules.     

Design of topological indices: computer-oriented approach

A novel method is suggested for constructing topological indices (TIs) of molecular graphs which models human logic. This method is described in terms of a block scheme, consisting of the mutually connected elementary blocks. In each block the simple transformations of a molecular graph are fulfilled. A variant of the transformation is selected from the list of possible variants. Every TI is obtained as a result of the sequential execution of a number of operations, corresponding to some ‘walk’ on the block scheme. This walk can be selected both randomly and by the investigator. The suggested method can serve as a basis for the development of the respective computer program which may be used for the automatic construction of any number of TIs of differing nature. By this process one can also obtain the TIs that are unlikely to be constructed manually, due to their complexity. The set of obtained TIs may be used for building the structure–property models. In the case of an unsatisfactory result the obtained set of TIs may be changed using the described generator of TIs. A number of examples of application of the suggested approach for the building QSAR/QSPR models is given.     

Four new topological indices based on the molecular path code

The sequence of all paths pi of lengths i = 1 to the maximum possible length in a hydrogen-depleted molecular graph (which sequence is also called the molecular path code) contains significant information on the molecular topology, and as such it is a reasonable choice to be selected as the basis of topological indices (TIs). Four new (or five partly new) TIs with progressively improved performance (judged by correctly reflecting branching, centricity, and cyclicity of graphs, ordering of alkanes, and low degeneracy) have been explored. (i) By summing the squares of all numbers in the sequence one obtains Sigmaipi(2), and by dividing this sum by one plus the cyclomatic number, a Quadratic TI is obtained: Q = Sigmaipi(2)/(mu+1). (ii) On summing the Square roots of all numbers in the sequence one obtains Sigmaipi(1/2), and by dividing this sum by one plus the cyclomatic number, the TI denoted by S is obtained: S = Sigmaipi(1/2)/(mu+1). (iii) On dividing terms in this sum by the corresponding topological distances, one obtains the Distance-reduced index D = Sigmai{pi(1/2)/[i(mu+1)]}. Two similar formulas define the next two indices, the first one with no square roots: (iv) distance-Attenuated index: A = Sigmai{pi/[i(mu + 1)]}; and (v) the last TI with two square roots: Path-count index: P = Sigmai{pi(1/2)/[i(1/2)(mu + 1)]}. These five TIs are compared for their degeneracy, ordering of alkanes, and performance in QSPR (for all alkanes with 3-12 carbon atoms and for all possible chemical cyclic or acyclic graphs with 4-6 carbon atoms) in correlations with six physical properties and one chemical property.     

Quantum-chemical calculations of carbon-isotope fractionation in CO2(g), aqueous carbonate species, and carbonate minerals

Quantum chemical calculations on large supermolecular carbonate-water and carbonate mineral clusters are used to predict equilibrium constants for 13,12C-isotope-exchange reactions between CO2(g), aqueous carbonate species, and the common carbonate minerals. For the aqueous species, we evaluate the influence of the size and conformational variability of the solvation shell, the exchange-correlation functional, and the basis set. The choice of exchange-correlation functional (PBE vs B3LYP), the basis set (6-31G* vs aug-cc-pVDZ), and solvation shell size (first shell only vs first shell and a partial second shell) each produce changes of approximately 5-10 per mil in the reduced partition function ratio. Conformational variability gives rise to a standard error of approximately 0.5 per mil using approximately 10 solute-solvent conformations. The best results are obtained with the B3LYP/aug-cc-pVDZ combination, but because the improvements in the basis set and exchange correlation functional drive the reduced partition function ratios in opposite directions, reasonably good results are also obtained with the PBE/6-31G* combination. To construct molecular clusters representative of mineral environments, a new method is introduced on the basis of conservation of Pauling bond strength. Using these clusters as models for minerals, calculations of mineral-gas and mineral-aqueous carbon-isotope fractionation factors, are in good agreement with experimental measurements. Carbon-isotope fractionation factors for gas, aqueous, and mineral phases are thus integrated into a single theoretical/computational framework.     

Cluster models and ab initio calculations of (19)F NMR isotropic chemical shifts for inorganic fluorides

(19)F NMR isotropic chemical shift (delta(iso)) calculations are performed in crystallized compounds using the GIAO method with the B3LYP hybrid functional at DFT level. Clusters centered on the studied fluorine atoms mimic the crystalline structures. The 6-311+G(d) basis set is chosen for the central fluorine atom, and the LanL2DZ basis set for the others. The metal atoms are described by the 3-21G(2d) basis set or, when not available, by the CRENBL basis set with the corresponding ECP, and augmented with 2d polarization functions when existing. First, for high-symmetry systems (MF, MF(2), and MF(3) compounds), a systematization of the cluster building up from coordination spheres is proposed, generalized to fluoroperovskites and fluoroaluminates KAlF(4) and RbAlF(4). When applied to rather low symmetry systems such as barium fluorometalates BaMgF(4), BaZnF(4), and Ba(2)ZnF(6), the definition of the coordination spheres is far from easy. Then, for structures built up from a MF(6) octahedron network, we may define different "starting clusters": [FM(2)F(8)] for the shared fluorine atoms, [FMF(4)] for the unshared ones, and [FBa(4)](7+) for the "free" ones. Analogous "starting clusters" are then tested on compounds from the NaF-AlF(3), BaF(2)-AlF(3), and CaF(2)-AlF(3) binary systems and for alpha-BaCaAlF(7) that are also built up from a MF(6) octahedron network. For each of these corresponding fluorine sites, delta(iso) values are calculated with the "starting clusters" and several larger clusters and compared to the experimental delta(iso) values. For the barium-containing clusters, the RMS deviation is equal to 51 ppm. It is suggested that this result may be related to the poor quality of the barium basis sets for which no polarization functions are available for the moment. In total, chemical shifts were calculated for 122 fluorine sites, in a various range of compounds. For the clusters without barium, the ab initio method leads to a RMS equal to 22 ppm, which is a quite nice result keeping in mind that the (19)F chemical shift range is larger than 200 ppm.     

Density functional studies on hydrogen‐bonded clusters of hydrogen halides and the interaction on halide anions

Density functional theory (DFT) calculations have been performed to study the structures and stability of X^?·(HX)_n=2–5 clusters where X = F, Cl, Br at B3LYP/6‐311++G** level of theory. The presence of halide ions in these clusters disintegrates the hydrogen halide clusters. All the hydrogen halides are then hydrogen bonded to the centrally placed halide ions, thereby forming multiple hydrogen bonds. The interaction energies have been corrected for the basis set superposition error (BSSE) using Boy's counterpoise correction method. Evidence for the destruction of hydrogen bonds in hydrogen halide clusters due to the presence of halide ions is further obtained from topological analysis and natural bond orbital analysis. The chemical hardness and chemical potential have been calculated for all the anion clusters. The above analysis reveals that hydrogen bonding in these systems is not an essentially electrostatic interaction. The nature of the stabilization interactions operative in these multiple hydrogen‐bonded clusters has been explained in terms of many‐body contribution to interaction energies. From these studies, an attempt has been made to understand the nature of the molecular properties resulting from different electronegativities of the halogens. © 2005 Wiley Periodicals, Inc. Int J Quantum Chem, 2005     

Comparative study of topological indices of macro/supramolecular RNA complex networks

RNA function annotation is often based on alignment to a previously studied template. In contrast to the study of proteins, there are not many alignment-free methods to predict RNA functions if alignment fails. The use of topological indices (TIs) of RNA complex networks (CNs) to find quantitative structure-activity relationships (QSAR) may be an alternative to incorporate secondary structure or sequence-to-sequence similarity. Here, we introduce new QSAR-like techniques using RNA macromolecular CNs (mmCNs), where nodes are nucleotides, or RNA supramolecular CNs (smCNs), where nodes are RNA sequences. We studied a data set of 198 sequences including 18S-rRNAs (important phylogenetic molecular biomarkers). We constructed three types of RNA mmCNs: sequence-linear (SL), Cartesian-lattice (CL), and sequence-folding CNs (SF-CNs) and two smCNs: sequence-sequence disagreement CN (SSD) and sequence-sequence similarity (SSS-smCN). We reported the first comparative QSAR study with all these CIs and CNs, which includes: (i) spectral moments ( ( i )micro d ( w)) of SL-mmCNs (accuracy = 75.3%), (ii) electrostatic CIs (xi d ) of CL-mmCNs (>90%), (iii) thermodynamic parameters (Delta G, Delta H, Delta S, and T m) of SF-mmCNs (64.7%), (iv) disagreement-distribution moments ( M k ) of the SSD-smCN (79.3%), and (v) node centralities of the SSD-smCN (78.0%). Furthermore, we reported the experimental isolation of a new RNA sequence from Psidum guajava leaf tissue and its QSAR and BLAST prediction to illustrate the practical use of these methods. We also investigated the use of these CNs to explore rRNA diversity on bacteria, plants, and parasites from the Dactylogyrus genus. The HPL-mmCNs model was the best of all found. All the CNs and TIs, except SF-mmCNs, were introduced here by the first time for the QSAR study of RNA, which allowed a comparative study for RNA classification.     

Stability, structural, and magnetic phase diagrams of ternary ferromagnetic 3d-transition-metal clusters with five and six atoms

We report a theoretical investigation of free-standing Fe(x)Co(y)Ni(z) ternary clusters with x + y + z = 5 and 6. Our study is performed within density functional theory as implemented in the GAUSSIAN 03 set of programs and with the BPW91/SDD level of theory. We analyze the geometries, chemical order, local and total magnetic moments, binding energies, excess energies, and second difference in the energy in the whole range of composition, from which structural, magnetic, and stability phase diagrams are predicted for these cluster sizes. We determine the optimal stoichiometries for these clusters as regards the maximum total magnetic moment and stability.     

Thermodynamic equilibrium compositions, structures, and reaction energies of Pt(x)O(y) (x = 1-3) clusters predicted from first principles

As synthetic nanocatalysis strives to create and apply well-defined catalytic centers containing as few as a handful of active metal atoms, it becomes particularly important to understand the structures, compositions, and reactivity of small metal clusters as a function of size and chemical environment. As a part of our effort to better understand the oxidation chemistry of Pt clusters, we present here a comprehensive set of density functional theory simulations combined with thermodynamic modeling that allow us to map out the T-p(O)2 phase diagrams and predict the oxygen affinity of Pt(x)O(y) clusters, x = 1-3. We find that the Pt clusters have a much stronger tendency to form oxides than does the bulk metal, that these oxides persist over a wide range of oxygen chemical potentials, and that the most stable cluster stoichiometry varies with size and may differ from the stoichiometry of the stable bulk oxide in the same environment. Further, the facility with which the clusters are reduced depends both on size and on composition. These models provide a systematic framework for understanding the compositions and energies of redox reactions of discrete metal clusters of interest in supported and gas-phase nanocatalysis.     

GTI-space: the space of generalized topological indices

A new extension of the generalized topological indices (GTI) approach is carried out to represent “simple” and “composite” topological indices (TIs) in an unified way. This approach defines a GTI-space from which both simple and composite TIs represent particular subspaces. Accordingly, simple TIs such as Wiener, Balaban, Zagreb, Harary and Randić connectivity indices are expressed by means of the same GTI representation introduced for composite TIs such as hyper-Wiener, molecular topological index (MTI), Gutman index and reverse MTI. Using GTI-space approach we easily identify mathematical relations between some composite and simple indices, such as the relationship between hyper-Wiener and Wiener index and the relation between MTI and first Zagreb index. The relation of the GTI-space with the sub-structural cluster expansion of property/activity is also analysed and some routes for the applications of this approach to QSPR/QSAR are also given.     

Methanol in its own gravy. A PCM study for simulation of vibrational spectra

For studying both hydrogen bond and dipole-dipole interactions between methanol molecules (self-association) the geometry of clusters of increasing numbers of methanol molecules (n = 1,2,3) were optimized and also their vibrational frequencies were calculated with quantum chemical methods. Beside these B3LYP/6-311G** calculations, PCM calculations were also done for all systems with PCM at the same quantum chemical method and basis set, for considering the effect of the liquid continuum on the cluster properties. Comparing the results, the measured and calculated infrared spectra are in good accordance.     

Vibrational analysis of I2*- x nCO2 clusters (n = 1-10): a first principle study on microsolvation

Structure, stability, and vibrational IR and Raman spectra of I(2)(*-) x nCO(2) clusters (n = 1-10) are reported based on first-principle electronic structure calculations. Several close-lying minimum energy structures are predicted for these solvated clusters following the quasi Newton-Raphson procedure of geometry optimization. Search strategy based on Monte-Carlo simulated annealing is also applied to find out the global minimum energy structures of these clusters. Successive addition of solvent CO(2) molecules to the negatively charged diatomic solute, I(2)(*-), is fairly symmetrical. Energy parameters of these solvated clusters are calculated following second-order Moller-Plesset perturbation (MP2) as well as coupled cluster theory with 6-311+G(d) set of basis function (I atom is treated with 6-311G(d) set of basis function). The excess electron in these solvated clusters is observed to be localized mainly over the two I atoms. Average interaction energy between the anionic solute, I(2)(*-), and a solvent CO(2) molecule is approximately 129 meV in I(2)(*-) x nCO(2) clusters, and the average interaction energy between two solvent CO(2) molecules is approximately 85 meV in the case of neutral (CO(2))(n) clusters at MP2 level of theory. IR spectra show similar features in all these solvated clusters, depicting a strong band at approximately 2330 cm(-1) for C-O stretching and a weak band at approximately 650 cm(-1) for CO(2) bending modes. Degeneracy of the bending mode of a free solvent CO(2) unit gets lifted when it interacts with the charged solute I(2)(*-) to form a molecular cluster because of the change in structure of solvent CO(2) units. The vibrational band at the bending region of CO(2) in the Raman spectra of these anionic clusters shows a characteristic feature for the formation of I(2)(*-) x nCO(2) clusters showing a Raman band at approximately 650 cm(-1).