Automated structure elucidation of organic molecules from (13)c NMR spectra using genetic algorithms and neural networks.

The automated structure elucidation of organic molecules from experimentally obtained properties is extended by an entirely new approach. A genetic algorithm is implemented that uses molecular constitution structures as individuals. With this approach, the structure of organic molecules can be optimized to meet experimental criteria, if in addition a fast and accurate method for the prediction of the used physical or chemical features is available. This is demonstrated using (13)C NMR spectrum as readily obtainable information. (13)C NMR chemical shift, intensity, and multiplicity information is available from (13)C NMR DEPT spectra. By means of artificial neural networks a fast and accurate method for calculating the (13)C NMR spectrum of the generated structures exists. The approach is limited by the size of the constitutional space that has to be searched and by the accuracy of the shift prediction for the unknown substance. The method is implemented and tested successfully for organic molecules with up to 20 non-hydrogen atoms.     

Electronic structure and spectra of organic molecules

Calculations of the singlet-singlet ( * type) transitions of a series of 2,6-disubstituted purines are presented. The Pariser-Parr-Pople method is used throughout and the agreement with experiment is good. The effects of the 2-X and 6-X substituents on the purine spectrum are discussed.For Part VIII (SCF MO CI Calculations for Hypoxanthine and Guanine Isomers) see Kwiatkowski, J. S.: Acta physica polon. 36 (1969), in press.     

Electronic structure and spectra of organic molecules

The Pariser-Parr-Pople method is used to predict the singlet-singlet ( ^* type). Electronic transition energies for the anions of hydroxy derivatives of benzene, pyridine and pyrimidine. The core integral for the -O^– substituent was treated as an empirical parameter and its value was chosen to fit results for the phenoxide ion and then applied for the compounds with the -O^– substituents. The results of the calculations for molecules with the -O^– groups are compared with those of the corresponding molecules containing the -OH and -NH₂ substituents as well as with the experimental data. No serious deviations between theoretical and experimental data other than those which could be ascribed to steric hindrances occuring in the molecules were found.

---

Zusammenfassung Die Pariser-Parr-Pople-Methode wird benutzt, um die Singlett-Singlett-Übergangsenergien (- ^*-Typ) von Benzol-, Pyridin- und Pyrimidinderivaten zu berechnen, die OH-Gruppen enthalten. Das Rumpfintegral des -O^–-Substituenten wurde wie ein empirischer Parameter behandelt. Sein Wert wurde so gewählt, daß er die Meßergebnisse für das Phenolat-Anion richtig wiedergibt. Die Resultate der Berechnungen für Moleküle mit der -O^–-Gruppe werden mit denjenigen für die entsprechenden Moleküle mit OH- und NH₂-Substituenten als auch mit den experimentellen Daten verglichen. Es werden keine ernsten Abweihungen zwischen theoretischen und experimentellen Daten gefunden, außer solchen, die einer sterischen Hinderung zugeschrieben werden können.

---

Résumé La méthode de Pariser, Parr et Pople est utilisée pour prédire les transitions électroniques ( ^*) pour les formes anioniques des derivés hydroxy du benzène, pyridine et pyrimidine. La valeur d'integrale de coeur associée du substituant -O^– a été choisie de façon à reproduire les positions des bandes d'absorption de forme anionique du phenol et après elle a été appliquée pour les composés envisagés. Les resultats des calculs pour l'anions sont mis en comparaison avec les resultats obtenus pour les correspondantes derivés aminés et hydroxy et avec les donnés experimentales. Peu de déviations ont été obtenus, en dehors de ceux que l'on peut attribuer à des encombres steriques.

---

---

Based in part on a M. Sci. thesis by M. Berndt, N. Copernikus University, Toru, 1969 (unpublished, in Polish).     

Deriving the 3D structure of organic molecules from their infrared spectra

The representation of the 3D structure of a molecule by a radial distribution function (RDF) code is described. The use of the RDF code for the simulation of an infrared spectrum by a counterpropagation (CPG) neural network is shown. Furthermore, a CPG network can also be operated in reverse mode: on input of an infrared spectrum an RDF code is obtained for which a 3D structure can be searched in a database. An empirical modelling process is used to refine this 3D structure to obtain a three-dimensional model of the molecular structure that corresponds to the infrared spectrum.     

Electronic structure and spectra of organic molecules

The Pariser-Parr-Pople method has been used to calculate the -electronic structures and spectra of 6-fluoro, 6-hydroxy, 6-methoxy, 6-amino, 6-methylamino, 6-dimethylaminopurine and their parent molecule purine. The purine and adenine (6-aminopurine) absorption spectra (singlet-singlet ^* type transitions) in particular are discussed in greater detail. The correlation among the absorption bands of the molecules in question and their ground state properties are also discussed. The theoretical results for singlet-singlet transition energies agree fairly well with the experimental data.

---

Zusammenfassung Mit der Methode von Pariser, Parr und Pople werden Struktur und Spektren der -Elektronensysteme von Purin sowie 6-F-, 6-OH-, 6-OCH₃-, 6-NH₂-, 6-NHCH₃- und 6-N(CH₃)₂-Purin berechnet, wobei besonderes Gewicht auf spektrale Übergänge von Purin und Adenin gelegt wurde. Ferner wurde die Korrelation der Purin-Spektren mit den Eigenschaften des Grundzustandes untersucht. Die theoretischen Ergebnisse für Singulettübergänge stimmen ziemlich gut mit den experimentellen Daten überein.

---

Résumé A l'aide de la méthode Pariser-Parr-Pople nous avons calculé les structures et les spectres électroniques d'une série de purines (purine, 6-F, 6-OH, 6-OCH₃, 6-NH₂, 6-NHCH₃ et 6-N(CH₃)₂-purine). L'accent a été mis en particulier sur les bandes d'absorption pour la purine et l'adénine. La corrélation entre les spectres d'absorption électronique et l'état fondamental de purines est étudiée. Les résultats théoriques concernant les énergies des transitions singulet-singulet sont en très bon accord avec les données expérimentales.

---

---

Submitted in honour of the seventieth birthday of Professor Aleksander Jabloski.     

Recent advances in the structure elucidation of small organic molecules by the LSD software

The LSD software proposes the structures of small organic molecules that fit with structural constraints from 1D and 2D NMR spectroscopy. Its initial design introduced limits that needed to be eliminated to extend its scope and help its users choose the most likely structure among those proposed. The LSD software code has been improved, so that it recognizes a wider set of atom types to build molecules. More flexibility has been given in the interpretation of 2D NMR data, including the automatic detection of very long‐range correlations. A program named pyLSD was written to deal with problems in which atom types are ambiguously defined. It also provides a ¹³C NMR chemical shift‐based solution ranking algorithm. PyLSD was able to propose the correct structure of hexacyclinol, a natural product whose structure determination has been highly controversal. The solution was ranked first within a list of ten structures that were produced by pyLSD from the literature NMR data. The structure of an aporphin natural product was determined by pyLSD, taking advantage of the possibility of handling electrically charged atoms. The structure generation of the insect antifeedant azadirachtin by LSD was reinvestigated by pyLSD, considering that three ¹³C resonances did not lead to univocal hybridization states. Copyright © 2013 John Wiley & Sons, Ltd.     

An expert system for inferring structures of organic compounds from their mass spectra.

Development of an 'expert system' for elucidation of structures of acyclic organic compounds is described. An expert system is a computer program that embodies some of the heuristic problem-solving knowledge of human experts so that it can effectively be used as an aid to decision making. The expert system described in this paper is intended to assist a chemist in arriving at plausible structures, the input data being the mass spectrum, molecular formula and presence (if known) of functional groups. The program generates chemically possible structures for the given molecular formula and can use, where available, the constraints imposed by the mass spectrum of the compound and by any known functional groups. The program makes use of a new algorithm for obtaining a canonical representation of structures and a new heuristic for incorporating constraints of the mass spectrum. This work constitutes a case study of the application of artificial intelligence techniques in chemistry and the material presented highlights this motivation.     

Data selection for structure generation in structure elucidation systems using molecular spectroscopy databases

Data selection techniques are considered for structure generation in structure elucidation systems using molecular spectroscopy databases. The starting data are sets of microfragments and connected structural fragments obtained from computer-aided analysis of mass, IR, and NMR spectra. Selection of fragments that do not isomorphically fit in larger fragments mostly leads to correct results with fewer output structures anddemands less computer time. Novosibirsk Institute of Organic Chemistry, Siberian Branch, Russian Academy of Sciences. Translated fromZhurnal Strukturnoi Khimii, Vol. 35, No. 2, pp. 46–53, March–April, 1994. Translated by L. Smolina     

Computer methods of structure elucidation by1H and13C NMR spectra using factographic databanks

Computer methods have been developed for structure elucidation by ¹ H and ¹³ C NMR spectra using factographic databanks containing spectral and structural data of many organic compounds (44,000 for ¹ H NMR and 27,000 for ¹³ C NMR). Information about the structure of compounds under study is obtained from the analysis of reference structures whose spectra are the best matches of the query spectra. This procedure identifies linked fragments composed of the nodes assigned to signals of the query spectrum when comparing it with reference spectra, and of inactive nodes separated from the former by one bond at most. Using many examples, it has been shown that this approach allows one to determine sufficiently large structural fragments of unknowns both by ¹ H and ¹³ C NMR spectra. The most reliable structures are obtained from the combined analysis of ¹ H and ¹³ C NMR spectral search results. Thus the relative occurrences of correct fragments among the n-first (n=1, 3, 5) fragments of the ranked list of candidates, output to the user, are 80, 93, and 96%, respectively.Institute of Organic Chemistry, Siberian Branch, Russian Academy of Sciences. Translated fromZhurnal Strukturnoi Khimii, Vol. 34, No. 3, pp. 51–60, May–June 1993.Translated by L. Smolina     

Computer-assisted structure elucidation of organic compounds III: Automatic fragment generation from13C-NMR spectra

The interpretation of carbon-NMR spectra is mainly based on the comparison with suitable reference data taken from literature. The whole information contents of¹³C-NMR spectra cannot be utilized by manual interpretation. Therefore a network of interactive computer programs has been developed, which simulates the strategy of the spectroscopist in generating structural fragments from the spectral data. The most important knowledge source for this process is a carbon-NMR data base containing some 17,500 spectra. Structural fragments are generated automatically from this data file and assembled by a model builder to complete chemical structures using constraints derived from the spectral data. A comparison of the experimental carbon-NMR spectrum with the estimated ones allows the generation of a sorted hitlist.For part II see: H. Kalchhauser, W. Robien,J. Chem. Inf. Comput. Sci. 1985,25, 103.     

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.     

New computer-assisted methods for the elucidation of molecular structure from 2-D spectra

General principles of the construction of expert systems for the elucidation of the structure of molecules from their spectra were considered. The principal attention was focused on systems based on the use of 2-D NMR spectra. The structural information extracted from 2D NMR spectra was characterized, and the strategy was outlined for structure elucidation under the conditions when the analyzed spectrostructural information is incomplete, fuzzy, and contradictory. The most advanced expert system ACD/Structure Elucidator, which is capable of determining the structure and stereochemistry of large molecules, in particular, those typical in the chemistry of natural compounds, is described as an example.     

A computer search system for chemical structure elucidation based on low-resolution mass spectra

A computerized search system which employs the data on the masses and relative abundances of spectral peaks and primary neutral losses is designed for computer elucidation of chemical structures. Recognition of structural fragments is based on analysis of the structures of reference compounds selected as best matches to the mass spectrum of the compound under investigation. Tests of the system on 67 “unknowns” show that the probability of recognizing a large structural fragment lies in the interval 60–80%, depending on the fragment size (100–50% of molecular weight), and that the reliability of the corresponding structural conclusion is 98%. An approach to automatic selection of the substructure common to all or several of the selected compounds is discussed.     

Structure elucidation from 2D NMR spectra using the StrucEluc expert system: detection and removal of contradictions in the data

The elucidation of chemical structures from 2D NMR data commonly utilizes a combination of COSY, HMQC/HSQC, and HMBC data. Generally COSY connectivities are assumed to mostly describe the separation of protons that are separated by 1 skeletal bond (3JHH), while HMBC connectivities represent protons separated from carbon atoms by 1 to 2 skeletal bonds (2JCH and 3JCH). Obviously COSY and HMBC connectivities of lengths greater than those described have been detected. Though experimental techniques have recently been described to aid in the identification of the nature of the couplings the detection of whether a coupling is 2-bond or greater still remains a challenge in most laboratories. In the StrucEluc software system the common lengths of the connectivities, 1-bond for COSY and 1- or 2-bond for HMBC, derived from 2D NMR data are set as the default. Therefore, in the presence of any extended connectivities contradictions can appear in the 2D NMR data. In this article, algorithmic methods for the detection and removal of contradictions in 2D NMR data that have been developed in support of StrucEluc are described. The methods are based on the analysis of molecular connectivity diagrams, MCDs. These methods have been implemented in the StrucEluc system and tested by solving 50 structural problems with 2D NMR spectral data containing contradictions. The presence of contradictions was detected by the algorithm in 90% of the cases, and the contradictions were automatically removed in approximately 50% of the problems. A method of "fuzzy" structure generation in the presence of contradictions has been suggested and successfully tested in this work. This work will demonstrate examples of the application of developed methods to a number of structural problems.     

Classification of organic nitrogen -containing compounds based on the electronic structure of their molecules

Conclusions It was shown that it is theoretically possible to classify nitrogen-containing compounds on the basis of the obtained quantum-chemical data concerning their electronic structure.Translated from Izvestiya Akademii Nauk SSSR, Seriya Khimicheskaya, No. 11, pp. 2611–2613, November, 1982.     

Electronic spectra of organic molecules and their interpretation—V Effect of terminal groups containing multiple bonds on the K-bands of conjugated systems

The investigation of the effect of terminal groups of absorbing (conjugated) systems on the corresponding K-bands is extended to groups containing multiple bonds. A consistent theoretical interpretation of all observations is given based on the consideration of the electronic perturbations in the excited state (due to Heisenberg resonance) and the view that organic molecules are built up of localised bonds and involve inductive electron-shifts only. The empirical comparative analysis of the experimental data shows conclusively that, in contrast to the conclusions arrived at by mathematical methods, the 2000 Å, and not the 1800 Å, band system of benzene corresponds to an excited state of the (approximate) symmetry ¹E_1μ (involving an allowed transition characteristic for all K-bands).     

Genius: a genetic algorithm for automated structure elucidation from 13C NMR spectra

The automated structure elucidation of organic molecules from experimentally obtained properties is extended by an entirely new approach. A genetic algorithm is implemented that uses molecular constitution structures as individuals. With this approach, the structure of organic molecules can be optimized to meet experimental criteria, if in addition a fast and accurate method for the prediction of the used physical or chemical features is available. This is demonstrated using 13C NMR spectrum as readily obtainable information. By means of artificial neural networks a fast and accurate method for calculating the 13C NMR spectrum of the generated structures exists. The method is implemented and tested successfully for organic molecules with up to 18 non-hydrogen atoms.     

Fine structure of the X-ray spectra of molecules K-absorption spectra of the CO and N2 molecules

Journal of Structural Chemistry -