THE SUMMARIES OF THE
PAPERS
-
G.N. Andreev, O. K. Argirov and P.N.
Penchev; Expert system for interpretation of infrared spectra. Analytica
Chimica Acta, 284, 131-136 (1993). Summary:
An expert system for the interpretation of infrared spectra EXPIRS was
created. The main features of EXPIRS are: hierarchical organization of
the characteristic groups, realized by frames; registration of the multiple
use of spectral bands; taking into account the solvent absorption and the
chemical inconsistencies; documenting the interpretation course and providing
explanations on request. The ten most important heuristics used by an expert
for interpretation of infrared spectra were formulated and tested with
EXPIRS.
-
P.N. Penchev, O.K. Argirov and G.N.
Andreev; Mass Spectra Classification According to Substructures and
Molecular Formula Using Artificial Neural Networks. Analytical
Laboratory, 3, 29-33 (1994). Summary:
Two artificial neural networks were designed to classify low-resolution
mass spectra of unknown compounds according to the presence or absence
of 16 substructures, and also to the molecular formula. The neural networks'
ability for predictions depends on the number of hidden neurons and on
the spectra preprocessing. To test the neural networks 140 mass spectra
of compounds with known structure were treated.
-
P.N. Penchev, A.N. Sohou and G.N.
Andreev; Description and Performance Analyses of An Infrared Libraray
Search System. Spectroscopy Letters, 29, 1513-1522
(1996). Summary: An infrared library search
system is described. The spectral library consists of 608 FT-IR spectra
represented with a data point every 4 cm-1 in the 3700-500 cm-1
range. Four different similarity measures for spectral search were implemented.
Performance analysis was carried out in order to estimate the ability of
the system to identify organic compounds on the basis of their IR spectra.
-
G.N. Andreev, B.A. Stamboliyska, P.N.
Penchev; Vibrational Spectra and Structure of 1,4-Dinitrobenzene
and its 15N Labelled Derivatives: Experimental and Ab Initio
Force Field Treatments. Spectrochimica Acta, Part A, 53,
811-818 (1997). Summary: Infrared and Raman
spectra in solid state and solution of 1,4-dinitrobenzene, 1,4-dinitrobenzene-15N,15N
and 1,4-dinitrobenzene-14N,15N have been studied
and their fundamental frequencies have been assigned using isotopic frequency
shifts and Raman depolarization ratios. Ab initio quantum chemical calculations
have been carried out for the three 1,4-dinitrobenzene isotopomers at the
3-21G, 6-31G and 6-31G** basis set levels and the computed vibrational
frequencies have been compared with the experimental ones. Best coincidence
was achieved with the frequencies calculated at the 6-31G level: the mean
deviation is 26.5 cm-1. The calculated isotopic frequency shifts,
induced by the 15N labelling, are in a very good accordance
with the measured ones. Except the C-H bond lengths, all the geometry parameters,
calculated for the isolated molecule, are in good agreement with the X-ray
data, obtained for the 1,4-dinitrobenzene monocrystal. The nitro groups
in the molecule hold charges of 0.390 e- each; there is practically
no nitro group / phenylene ring conjugation.
-
K. Varmuza, P. Penchev, F. Stancl,
W. Werther; Systematic Structure Elucidation of Organic Compounds by
Mass Spectra Classification. Journal of Molecular Structure,
408/409, 91-96 (1997). Summary:
Presence or absence of substructures is predicted from low resolution
mass spectra by multivariate classification methods. Classification results
are evaluated and then transformed to suitable structural restrictions
for isomer generation. An example with a compound C10H12O3 demonstrates
potential applications of this approach for systematic structure elucidation
of organic compounds.
-
K. Varmuza, P. Penchev, H. Scsibrany.
Maximum Common Substructures of Organic compounds Exhibiting Similar
Infrared Spectra. Journal of Chemical Information and Computer Sciences,
38, 420-427 (1998). Summary:
Information about the unknown chemical structure of an organic compound
can be obtained by comparing the infrared spectrum with the spectra of
a spectral library. The resulting hitlist contains compounds exhibiting
the most similar spectra. A method based on the maximum common substructure
concept has been developed for an automatic extraction of common structural
features from the hitlist structures. A set of substructures is derived
that are characteristic for the query structure. Results can been used
as structural restrictions in isomer generation.
-
K. Varmuza, P.N.
Penchev, H. Scsibrany; Large and Frequently Occurring Substructures
in Organic Compounds Obtained by Library Search of Infrared Spectra. Vibrational
Spectroscopy, 19, 407-412 (1999). Summary:
Comparing the infrared spectrum of a compound with unknown chemical structure
with the spectra of a library is a routinely used method to obtain information
about the unknown structure. The resulting hitlist contains compounds exhibiting
the most similar spectra. If the unknown is not contained in the library
a method based on the maximum common substructure concept can be applied
to extract common structural features from the hitlist structures. Result
is a set of substructures that are characteristic for the query structure.
This approach has been applied to infrared spectra from a series of model
compounds and has been compared with information obtained from mass spectra
by the same method. A complementary method for evaluating spectral hitlists
is principal component analysis of spectra and chemical structures.
-
P.N. Penchev, G.N. Andreev, K. Varmuza;
Automatic Classification of Infrared Spectra Using a Set of Improved
Expert-based Features. Analytica Chimica Acta, 388(1-2),
145-159 (1999). Summary: Three types
of spectral features derived from infrared peak tables were compared for
their ability to be used in automatic classification of infrared spectra.
Aim of classification was to provide information about presence or absence
of 20 chemical substructures in organic compounds. A new method has been
applied to improve spectral wavelength intervals as available from expert-knowledge.
The resulting set of features proved to be better than features derived
from the original intervals and better that features directly derived from
peak tables. The methods used for classification were linear discriminant
analysis and a back-propagation neural network; the latter gave a better
performance of the developed classifiers.
-
Plamen N. Penchev, Kurt Varmuza;
Characteristic substructures in sets of organic compounds with similar
infrared spectra.Computers & Chemistry, 25, 231–237,
(2001). Summary: A method based on the
determination of maximum common substructures is applied for the generation
of substructures which are characteristic for a given set of molecular
structures. The molecular structures are from hitlists obtained by spectral
library searches; the hitlists contain those reference compounds which
have infrared spectra most similar to that from the query compound. The
influences of various parameters of this method are investigated with the
aim to improve the relevance of the obtained substructures for the structure
of the query compound.
-
Kurt Varmuza, Nikolay T. Kochev and Plamen
N. Penchev; Evaluation of Hitlists from IR Library Searches by the
Concept of Maximum Common Substructures. Analytical Sciences,
17, i659-i662 (2001). Summary: Hitlists
from spectral similarity searches with IR spectra are used to generate
lists of substructures that are characteristic for the structures of the
unknowns. The applied method searches for maximum common substructures
in all pairs of hitlits structures and ranks these substructures by a newly
defined criterion. Examples demonstrate the advantage of the new ranking
criterion as well as potentials and limits of the method.
-
S. S. Wesolowski, M. L. Leininger, P.
N. Pentchev, and H. F. Schaefer III; Electron Affinities of the
DNA and RNA Bases. Journal of the American Chemical Society, 123,
4023-4028 (2001). Summary: Adiabatic electron
affinities (AEAs) for the DNA and RNA bases are predicted by using a range
of density functionals with a double-zeta plus polarization plus diffuse
(DZP++) basis set in an effort to bracket the true EAs. Although the AEAs
exhibit moderate fluctuations with respect to the choice of functional,
systematic trends show that the covalent uracil (U) and thymine (T) anions
are bound by 0.05-0.25 eV while the adenine (A) anion is clearly unbound.
The computed AEAs for cytosine (C) and guanine (G) oscillate between small
positive and negative values for the three most reliable functional combinations
(BP86, B3LYP, and BLYP), and it remains unclear if either covalent anion
is bound. AEAs with B3LYP/TZ2P++ single points are 0.19 (U), 0.16 (T),
0.07 (G), -0.02 (C), and -0.17 eV (A). Favorable comparisons are made to
experimental estimates extrapolated from photoelectron spectra data for
the complexes of the nucleobases with water. However, experimental values
scaled from liquid-phase reduction potentials are shown to overestimate
the AEAs by as much as 1.5 eV. Because the uracil and thymine covalent
EAs are in energy ranges near those ofv their dipole-bound counterparts,
preparation and precise experimental measurement of the thermodynamically
stable covalent anions may prove challenging.