Markus Meringer

y-Randomization and its variants in QSPR/QSAR

y-Randomization is a tool used in validation of QSPR/QSAR models, whereby the performance of the original model in data description (r2) is compared to that of models built for permuted (randomly shuffled) response, based on the original descriptor pool and the original model building procedure. We compared y-randomization and several variants thereof, using original response, permuted response, or random number pseudoresponse and original descriptors or random number pseudodescriptors, in the typical setting of multilinear regression (MLR) with descriptor selection. For each combination of number of observations (compounds), number of descriptors in the final model, and number of descriptors in the pool to select from, computer experiments using the same descriptor selection method result in two different mean highest random r2 values. A lower one is produced by y-randomization or a variant likewise based on the original descriptors, while a higher one is obtained from variants that use random number pseudodescriptors. The difference is due to the intercorrelation of real descriptors in the pool. We propose to compare an original models r2 to both of these whenever possible. The meaning of the three possible outcomes of such a double test is discussed. Often y-randomization is not available to a potential user of a model, due to the values of all descriptors in the pool for all compounds not being published. In such cases random number experiments as proposed here are still possible. The test was applied to several recently published MLR QSAR equations, and cases of failure were identified. Some progress also is reported toward the aim of obtaining the mean highest r2 of random pseudomodels by calculation rather than by tedious multiple simulations on random number variables.

Fast generation of regular graphs and construction of cages

A formula is developed for the number of congruence classes of 2-cell imbeddings of complete bipartite graphs in closed orientable surfaces.

Consensus structure elucidation combining GC/EI-MS, structure generation, and calculated properties.

This article explores consensus structure elucidation on the basis of GC/EI-MS, structure generation, and calculated properties for unknown compounds. Candidate structures were generated using the molecular formula and substructure information obtained from GC/EI-MS spectra. Calculated properties were then used to score candidates according to a consensus approach, rather than filtering or exclusion. Two mass spectral match calculations (MOLGEN-MS and MetFrag), retention behavior (Lee retention index/boiling point correlation, NIST Kovats retention index), octanol-water partitioning behavior (log K(ow)), and finally steric energy calculations were used to select candidates. A simple consensus scoring function was developed and tested on two unknown spectra detected in a mutagenic subfraction of a water sample from the Elbe River using GC/EI-MS. The top candidates proposed using the consensus scoring technique were purchased and confirmed analytically using GC/EI-MS and LC/MS/MS. Although the compounds identified were not responsible for the sample mutagenicity, the structure-generation-based identification for GC/EI-MS using calculated properties and consensus scoring was demonstrated to be applicable to real-world unknowns and suggests that the development of a similar strategy for multidimensional high-resolution MS could improve the outcomes of environmental and metabolomics studies.

MOLGEN-CID: A canonizer for molecules and graphs accessible through the internet

The MOLGEN Chemical Identifier MOLGEN-CID is a software module freely accessible via the Internet. For a molecule or graph entered in molfile format (2D) it produces, by a canonical renumbering procedure, a canonical molfile and a unique character string that is easily compared by computer to a similar string. The mode of operation of MOLGEN-CID is detailed and visualized with examples.

QSPR Using MOLGEN-QSPR: The Example of Haloalkane Boiling Points

MOLGEN-QSPR is a software newly developed for use in quantitative structure property relationships (QSPR) work. It allows to import, to manually edit, or to generate chemical structures, to detect duplicate structures, to import or to manually input property values, to calculate the values of a broad pool of molecular descriptors, to establish QSPR equations (models), and using such models to predict unknown property values. In connection with the molecule generator MOLGEN, MOLGEN-QSPR is able to predict property values for all compounds in a predetermined structure space (inverse QSPR). Some of the features of MOLGEN-QSPR are demonstrated on the example of haloalkane boiling points. The data basis used here is broader than in previous studies, and the models established are both more precise and simpler than those previously reported.

Small Molecule Identification with MOLGEN and Mass Spectrometry

This paper details the MOLGEN entries for the 2012 CASMI contest for small molecule identification to demonstrate structure elucidation using structure generation approaches. Different MOLGEN programs were used for different categories, including MOLGEN–MS/MS for Category 1, MOLGEN 3.5 and 5.0 for Category 2 and MOLGEN–MS for Categories 3 and 4. A greater focus is given to Categories 1 and 2, as most CASMI participants entered these categories. The settings used and the reasons behind them are described in detail, while various evaluations are used to put these results into perspective. As one author was also an organiser of CASMI, these submissions were not part of the official CASMI competition, but this paper provides an insight into how unknown identification could be performed using structure generation approaches. The approaches are semi-automated (category dependent) and benefit greatly from user experience. Thus, the results presented and discussed here may be better than those an inexperienced user could obtain with MOLGEN programs.

Beyond Terrestrial Biology: Charting the Chemical Universe of α-Amino Acid Structures

α-Amino acids are fundamental to biochemistry as the monomeric building blocks with which cells construct proteins according to genetic instructions. However, the 20 amino acids of the standard genetic code represent a tiny fraction of the number of α-amino acid chemical structures that could plausibly play such a role, both from the perspective of natural processes by which life emerged and evolved, and from the perspective of human-engineered genetically coded proteins. Until now, efforts to describe the structures comprising this broader set, or even estimate their number, have been hampered by the complex combinatorial properties of organic molecules. Here, we use computer software based on graph theory and constructive combinatorics in order to conduct an efficient and exhaustive search of the chemical structures implied by two careful and precise definitions of the α-amino acids relevant to coded biological proteins. Our results include two virtual libraries of α-amino acid structures corresponding to these different approaches, comprising 121 044 and 3 846 structures, respectively, and suggest a simple approach to exploring much larger, as yet uncomputed, libraries of interest.

Extraordinarily Adaptive Properties of the Genetically Encoded Amino Acids

Using novel advances in computational chemistry, we demonstrate that the set of 20 genetically encoded amino acids, used nearly universally to construct all coded terrestrial proteins, has been highly influenced by natural selection. We defined an adaptive set of amino acids as one whose members thoroughly cover relevant physico-chemical properties, or “chemistry space.” Using this metric, we compared the encoded amino acid alphabet to random sets of amino acids. These random sets were drawn from a computationally generated compound library containing 1913 alternative amino acids that lie within the molecular weight range of the encoded amino acids. Sets that cover chemistry space better than the genetically encoded alphabet are extremely rare and energetically costly. Further analysis of more adaptive sets reveals common features and anomalies, and we explore their implications for synthetic biology. We present these computations as evidence that the set of 20 amino acids found within the standard genetic code is the result of considerable natural selection. The amino acids used for constructing coded proteins may represent a largely global optimum, such that any aqueous biochemistry would use a very similar set.

QSPR Using MOLGEN-QSPR: The challenge of fluoroalkane boiling points

By means of the new software MOLGEN-QSPR, a multilinear regression model for the boiling points of lower fluoroalkanes is established. The model is based exclusively on simple descriptors derived directly from molecular structure and nevertheless describes a broader set of data more precisely than previous attempts that used either more demanding (quantum chemical) descriptors or more demanding (nonlinear) statistical methods such as neural networks. The models internal consistency was confirmed by leave-one-out cross-validation. The model was used to predict all unknown boiling points of fluorobutanes, and the quality of predictions was estimated by means of comparison with boiling point predictions for fluoropentanes.

MOLGEN 5.0, a Molecular Structure Generator

MOLGEN 5.x combines the efficiency of the molecular generator MOLGEN 3.5 and the flexibility of MOLGEN 4.x. To achieve this, the software was reimplemented based on a totally new concept. The most visible new features are fuzzy molecular formula input and explicit use of atom state patterns. We here describe the first version MOLGEN 5.0 of this new series.