380
P. M. Vassiliev et al.
8. the final estimate of compound activity should be made by generalizing the prediction results obtained from different methods.
These principles underlie the means for processing information about chemical
structure and biological properties that are discussed below.
12.2.2 QL Language for Fragment Coding of Chemical
Compound Structure
QSAR Language is a specialized multidescriptor, hierarchical, multilevel language
for the description of the structure of a chemical compound with fragmental substructural notation [107, 110]. It is the working language of IT Microcosm and
provides an adequate representation of a full molecular graph.
Substructural descriptors are the key elements of the language; a substructural
descriptor is a chemical structure fragment of a varying degree of complexity, isolated from the structural formula of a compound according to the rules and postulates of the QL language. The simplest fragments are called elementary descriptors;
they constitute the QL alphabet. All of the other descriptors are a combination of
several elementary descriptors; in relation to the elementary descriptors, they are
composite descriptors. Eleven descriptor types of the first four ranks would be sufficient for an unambiguous representation of a molecular graph in QL.
The QL alphabet (a list of first-rank descriptors) is postulated; it is defined by
three types of elementary descriptors.
A Structural Descriptor (SD) is a fragment of a compound structure with a sufficiently labile electron system constituting a non-hydrogen atom or a group of atoms with the immediate environment taken into consideration. The SD subalphabet
in QL is postulated; it contains 4352 types of descriptors for all of the elements in
the periodic table, including 378 types of heteroatom-containing SDs, 11 types of
carbon-containing SDs, 3963 types of cyclic SDs (of three classes—nonconjugated,
conjugated nonaromatic and aromatic rings smaller than 99 atoms) (Table 12.1). If a
SD shows chirality, its symbol bears an R or S in accordance with the Cahn-IngoldPrelog priority rules [82].
A Length Descriptor (LD) is the length in bonds of the shortest path along the
carbon chain between two SDs (values ranging from 1 to 99) or the number of total
atoms when one SD is superposed immediately over another (the occurrence index
has a negative value in this case). When looking for the shortest path, heteroatomic
SDs do not cross.
A Bond Descriptor (BD) characterizes the type of electron system in an LD and
reflects the types of bonds, their number and the presence of conjugation. The BD
subalphabet is postulated; it consists of 54 types of descriptors, where each one is
formed from a combination of 4 binary indices: the presence of one or several multiple bonds (p, P), aromatic (a, A) and noncovalent bonds (n, N), and conjugation
index (0, 1) (Table 12.2).
P. M. Vassiliev et al.
8. the final estimate of compound activity should be made by generalizing the prediction results obtained from different methods.
These principles underlie the means for processing information about chemical
structure and biological properties that are discussed below.
12.2.2 QL Language for Fragment Coding of Chemical
Compound Structure
QSAR Language is a specialized multidescriptor, hierarchical, multilevel language
for the description of the structure of a chemical compound with fragmental substructural notation [107, 110]. It is the working language of IT Microcosm and
provides an adequate representation of a full molecular graph.
Substructural descriptors are the key elements of the language; a substructural
descriptor is a chemical structure fragment of a varying degree of complexity, isolated from the structural formula of a compound according to the rules and postulates of the QL language. The simplest fragments are called elementary descriptors;
they constitute the QL alphabet. All of the other descriptors are a combination of
several elementary descriptors; in relation to the elementary descriptors, they are
composite descriptors. Eleven descriptor types of the first four ranks would be sufficient for an unambiguous representation of a molecular graph in QL.
The QL alphabet (a list of first-rank descriptors) is postulated; it is defined by
three types of elementary descriptors.
A Structural Descriptor (SD) is a fragment of a compound structure with a sufficiently labile electron system constituting a non-hydrogen atom or a group of atoms with the immediate environment taken into consideration. The SD subalphabet
in QL is postulated; it contains 4352 types of descriptors for all of the elements in
the periodic table, including 378 types of heteroatom-containing SDs, 11 types of
carbon-containing SDs, 3963 types of cyclic SDs (of three classes—nonconjugated,
conjugated nonaromatic and aromatic rings smaller than 99 atoms) (Table 12.1). If a
SD shows chirality, its symbol bears an R or S in accordance with the Cahn-IngoldPrelog priority rules [82].
A Length Descriptor (LD) is the length in bonds of the shortest path along the
carbon chain between two SDs (values ranging from 1 to 99) or the number of total
atoms when one SD is superposed immediately over another (the occurrence index
has a negative value in this case). When looking for the shortest path, heteroatomic
SDs do not cross.
A Bond Descriptor (BD) characterizes the type of electron system in an LD and
reflects the types of bonds, their number and the presence of conjugation. The BD
subalphabet is postulated; it consists of 54 types of descriptors, where each one is
formed from a combination of 4 binary indices: the presence of one or several multiple bonds (p, P), aromatic (a, A) and noncovalent bonds (n, N), and conjugation
index (0, 1) (Table 12.2).