Chemics—ube,a modified system of chemics

Two kinds of contraint have been added to the CHEMICS program system for structure elucidation of organic compounds. One is a limitation on the construction of small ring structures; the other is the introduction of a facility for input of appropriate information at the discretion of the user during the automated analyses of spectral data of unknown compounds. The computation processes for a couple of compounds by the new system are described.     

Computer-aided structure elucidation methods

Computer-aided structure elucidation methodology is discussed. Major processes involved in computer-aided structure elucidation systems are partial-structure elucidation, structure generation, and structure examination. For the three representative systems CONGEN, CASE, and CHEMICS, these processes are examined. There are four necessary conditions for automated chemical structure elucidation systems: reliability, width of application, ease of modification and portability.     

Computer-aided structure elucidation of organic compounds with the chemics system: Removal of Redundant Candidates by 13C-n.m.r. Prediction

A ¹³C-n.m.r. prediction module capable of removing inappropriate candidate structures given for an unknown compound based on the spectral data is introduced for the CHEMICS system. Given a set of candidate structures generated in the system, the routine may be used to prune off redundant candidates which have a predicted number of signals inconsistent with the observed number. It is shown that the addition of the examination module to the system makes structure elucidation by computer much more practical.     

Computer Assisted Structural Interpretation of Mass Spectral Data

An interactive system of programs has been developed to assist in structure elucidation based on mass spectral data. The program relies on algorithms for generating all the structural isomers that constitute alternative explanations of the observed data and it associates relative plausibility values with the different isomers. The structure assembly part of the program allows for the use of overlapping substructural components, such as substructures inferred from the appearance of particular ion patterns in the spectrum of an unknown compound. Mass spectrum interpretation procedures used with this structure assembly process could exploit any form of spectrum-substructure correlation scheme. In this work, the emphasis has been on the use of detailed and class specific spectrum-substructure correlations. Applications of the program are illustrated by means of an example analysis of the mass spectra of a variety of marine sterols.     

Structure generation by reduction: a new strategy for computer-assisted structure elucidation

A problem common to computer programs for structure elucidation is the efficient and prospective use of the input information to constrain the structure generation process. The input may consist of potentially overlapping substructure requirements and alternative substructure interpretations of spectral data. Other useful information may be structural features that must not be present in the output structures. All of these may interact in a complex manner that is impossible to determine by use of a bond-by-bond structure assembly algorithm. A new method is described called structure reduction. In contrast to structure assembly, this method begins with a set of all bonds and removes inconsistent bonds as structure generation progresses. This results in a more efficient use of the input information and the ability to use potentially overlapping required substructures. Several examples illustrate the application of our computer program COCOA, which uses this method to solve real-world structure elucidation problems.     

A computer program system — new chemics — for structure elucidation of organic compounds by spectral and other structural information

Computer-assisted structure elucidation is improved by the introduction of substructures selected by the user, in addition to analyses of the spectral data of an unknown organic compound. The substructure is called a 'macrocomponent' in the system. The macrocomponent which is input at will is authenticated by comparison with the set of components assembled by the automated analyses of the spectra before it is used for structure construction. It is shown that the introduction of the macrocomponent enhances the correctness and practicality of structure elucidation by computer.     

Spec2D: a structure elucidation system based on 1H NMR and H-H COSY spectra in organic chemistry

A system for structure elucidation based on proton NMR spectra has been developed. The system, named Spec2D (system for spectra from 2D-NMR), incorporates 1H NMR and H-H correlation spectroscopy (COSY) spectral information obtained from 2D-NMR experiments. 2D-NMR is important for the structure elucidation because it provides information about the relationships among differently situated protons in the structures of unknown compounds. The system uses the concepts of molecular graphs. The improved representation of substructures as well as several novel algorithms for structure generation have been devised to solve the combinatorial problem and to reduce the processing time. Spec2D consists of a knowledge base, an analysis module, and a candidate structure generator module. Spec2D proposes candidate structures from only 1H NMR and H-H COSY spectral information of an unknown compound without any 13C NMR spectral or structural information, such as molecular formulas. Spec2D has the capability to propose the "new" structure of an unknown compound, if the corresponding substructures are included in the knowledge base.     

Contribution to Structural Elucidation： Behaviours of Substructures Partially Defined from 2D NMR

Structural elucidation (automatic determination of the structure of a molecule from its spectra) is frequently hampered by com-binatorial explosion when trying to assemble the identified sub-structures. We devised a new method which can avoid this pit-fall by a systematic examination of allowed t3C chemical shifts ranges for all substructures chemically possible and combined with a progressive pruning thanks to neighbouring relationships appearing from 2D NMR. This method is explained by a de-tailed example.     

Practical aspects of liquid chromatography/electrospray tandem mass spectrometry for rapid identification of metabolites of a new antifungal agent SYN-2836 in dog urine

This report presents the structural elucidation of 12 urinary metabolites of SYN-2836, a new antifungal agent showing extensive metabolism in beagle dogs, using complementary liquid chromatography/tandem mass spectrometry (LC/MS/MS) methodologies. The 12 SYN-2836 metabolites were readily divided into four groups by considering that all three members of each group, although differing in masses, exhibited highly similar product ion mass spectra. This suggests that the metabolites within each group share a common major substructure. Therefore, all the grouped SYN-2836 metabolites were strategically identified by characterization of the major substructures followed by determination of the additional small substructures. This grouping strategy greatly facilitated the structural elucidation of these metabolites. Other strategies were also employed to achieve as rapid and unambiguous characterization of the SYN-2836 metabolites as possible.     

A framework for automated structure elucidation from routine NMR spectra

Methods to automate structure elucidation that can be applied broadly across chemical structure space have the potential to greatly accelerate chemical discovery. NMR spectroscopy is the most widely used and arguably the most powerful method for elucidating structures of organic molecules. Here we introduce a machine learning (ML) framework that provides a quantitative probabilistic ranking of the most likely structural connectivity of an unknown compound when given routine, experimental one dimensional ¹H and/or ¹³C NMR spectra. In particular, our ML-based algorithm takes input NMR spectra and (i) predicts the presence of specific substructures out of hundreds of substructures it has learned to identify; (ii) annotates the spectrum to label peaks with predicted substructures; and (iii) uses the substructures to construct candidate constitutional isomers and assign to them a probabilistic ranking. Using experimental spectra and molecular formulae for molecules containing up to 10 non-hydrogen atoms, the correct constitutional isomer was the highest-ranking prediction made by our model in 67.4% of the cases and one of the top-ten predictions in 95.8% of the cases. This advance will aid in solving the structure of unknown compounds, and thus further the development of automated structure elucidation tools that could enable the creation of fully autonomous reaction discovery platforms.

A machine learning model and graph generator were able to accurately predict for the presence of nearly 1000 substructures and the connectivity of small organic molecules from experimental 1D NMR data.     

Development of algorithms for automated elucidation of spectral feature/substructure relationships in tandem mass spectrometry

A pattern-recognition/artificial-intelligence program, referred to as MAPS (method for analyzing patterns in spectra), is described for the identification of relationships that exist between the presence of substructures in molecules and the characteristic features they produce in mass spectrometry (MS) and MS/MS data. The MAPS algorithm discovers these relationships by intelligent analysis of a data base of MS and MS/MS spectra. The relationships found are expressed as rules, which may then be used to identify characterized substructures in “unknowns”. No prior knowledge of fragmentation pathways or rearrangements is assumed in the rule-generation process. While MAPS currently uses MS and MS/MS data, the approach (and much of the software) is equally suited to multiple-stage mass spectrometric data.     

Exploring the limits of graph invariant- and spectrum-based discrimination of (sub)structures

The limits of a recently proposed computer method for finding all distinct substructures of a chemical structure are systematically explored within comprehensive graph samples which serve as supersets of the graphs corresponding to saturated hydrocarbons, both acyclic (up to n = 20) and (poly)cyclic (up to n = 10). Several pairs of smallest graphs and compounds are identified that cannot be distinguished using selected combinations of invariants such as combinations of Balaban's index J and graph matrix eigenvalues. As the most important result, it can now be stated that the computer program NIMSG, using J and distance eigenvalues, is safe within the domain of mono- through tetracyclic saturated hydrocarbon substructures up to n = 10 (oligocyclic decanes) and of all acyclic alkane substructures up to n = 19 (nonadecanes), i.e., it will not miss any of these substructures. For the regions surrounding this safe domain, upper limits are found for the numbers of substructures that may be lost in the worst case, and these are low. This taken together means that the computer program can be reasonably employed in chemistry whenever one is interested in finding the saturated hydrocarbon substructures. As to unsaturated and heteroatom containing substructures, there are reasons to conjecture that the method's resolving power for them is similar.     

Use of soft and hard ionization techniques for elucidation of unknown compounds by gas chromatography/time-of-flight mass spectrometry

Investigation of trace-level non-target compounds by gas chromatography/mass spectrometry (GC/MS) often is a challenging task that requires powerful software tools to detect the unknown components, to obtain the deconvoluted mass spectra, and to interpret the data if no acceptable library match is obtained. In this paper, the complementary use of electron ionization (EI) and chemical ionization (CI) is investigated in combination with GC/time-of-flight (TOF) MS for the elucidation of organic non-target (micro)contaminants in water samples. Based on accurate mass measurement of the molecular and fragment ions from the TOF MS, empirical formulae were calculated. Isotopic patterns, carbon number prediction filter and nitrogen rule were used to reduce the number of possible formulae. The candidate formulae were searched in databases to find possible chemical structures. Selection from possible structure candidates was achieved using information on substructures and observed neutral losses derived from the fragment ions. Four typical examples (bifenazate, boscalid, epoxiconazole, and fenhexamid) are used to illustrate the methodology applied and the various difficulties encountered in this process. Our results indicate that elucidation of unknowns cannot be achieved by following a standardized procedure, as both expertise and creativity are necessary in the process.     

Substructure, subgraph, and walk counts as measures of the complexity of graphs and molecules.

In discussions of unsaturated compounds represented by multigraphs it is necessary to distinguish between the notions of substructure and subgraph. Here the difference is explained and exemplified, and a computer program is introduced which for the first time is able to construct and count all substructures and subgraphs for a colored multigraph (a molecular compound which may contain unsaturation and heteroatoms). Construction of all substructures and subgraphs is computationally demanding; therefore, two alternatives are pointed out for the treatment of large sets of compounds: (i) Often it will suffice to consider counts of substructures/subgraphs up to a certain number of edges only, information which is provided by the program much more rapidly. (ii) It is shown that information equivalent to that gained from substructure or subgraph counts is often far more easily available using walk counts. Some problems and their consequences for substructure/subgraph/walk counts are discussed that arise from the models used in organic chemistry for certain compounds such as aromatics and from the necessity to express qualitative features of molecular structures numerically.     

Substructure Prediction from Infrared Spectra by Using Support Vector Machines

The potential of support vector machines （SVMs） for the substructure elucidation of infrared spectra have been investigated. The trained SVMs can identify the 16 substructures with high accuracy.     

Application of artificial intelligence for chemical inference. XXV. A computer program for automated empirical 13C NMR rule formation

A computer program which generates empirical rules associating ¹³C NMR shifts with local structural environments is described. The program uses a heuristic method to search for common structural features for those carbon atoms exhibiting similar shifts. Rules have been generated by our program from a combined set of acyclic amine and paraffin data. Examples of these rules are presented, and their performance as a tool for structure elucidation is examined.     

Computer generation of chemical structures from known fragments

The interactive generation of chemical structures from given fragments is described and discussed. It is implemented as a part of our expert system CARBON, based on C-13 NMR spectra. As it is designed, this program can also be a useful tool in the structure elucidation process when information on parts of the structure is obtained by other means (IR, mass and other spectrometries, chemical analysis, other relevant information). The topological characteristics of candidate fragments are first chosen interactively and then the elements are connected in all topologically possible ways. In the following step, the topological building blocks are substituted by chemical structural fragments resulting in a set of all chemical structures consistent with the input information.     

Fuzzy definition of molecular fragments in chemical structures

This paper presents a methodology for seeking the relationships between chemical substructures (molecular fragments) and spectral parameters using a computer collection data of molecular spectra. To establish the spectrum-structure correlations, the program has to search the chemical structure base in order to find compounds containing a given molecular fragment in the molecule. There exists no sole definition of a substructure, as it always depends on the type of problem dealt with. In the problem of structural identification, fuzzy definitions of substructures are applied, and their forms are imposed by the spectral methods used.