X e
i¼1
c
s
i À c
p
i
À
Á
¼ n
and
min m p ; m n
À
Á ¼ m
where
m p ¼
X e
i¼1
max c
s
i À c
p
i
À
Á ; 0
À
Á
m n ¼
X e
i¼1
min c
s
i À c
p
i
À
Á
; 0
À
Á
Here c
s
i , c
p
i and e have the same meaning as defined in the case of strong
matching while m p is the sum of vertex surplus and m n is the sum of vertex deficit
in the source distance distribution over the present distance distribution.
The procedure of cycle introduction, canonicalization and unique SMILES
notation generation is the same as done before.
Now, once the structures are generated using the methods described above, one
can use some user-defined parameters incorporated in the computer program to
restrict the number and size of the cycles to be created in the 2D structures. Few
other user-defined parameters, available in the program, may also be used to add
multiplicity of bonds (double and triple bonds) between pairs of vertices and other
hetero-atoms (e.g. nitrogen, oxygen, halogens) in order to get complete 2D structures of the compounds. The output of the generated structures may be saved in
SMILES notations and can be viewed using a molecular modelling software that is
capable of getting molecular structures from SMILES notation. Subsequently, the
activities of the generated structures may be predicted using the rule-based method
[18, 19] standardized for a biological endpoint of interest and can be prioritized and
screened from their MPS values. In this way, one may be able to screen some
potential bioactive compounds from the bigger set of combinatorially generated
molecular structures using topological distance information associated with
activity-related vertices present in the active compounds of a data set under consideration. It may be worth noting at this point that this newly developed method
[15] is essentially a molecular topology-based approach and activity prediction is
done using molecular graphs of the compounds where bond multiplicity and atom
types are not required. However, since bond multiplicity and atom types can be
introduced in the combinatorially generated topological structures using the options
available in the program and those structures can be saved in SMILES format, one
can always use these generated structures for any 2D and 3D drug design/discovery
applications.
Combinatorial Drug Discovery from Activity-Related Substructure …
89
i¼1
c
s
i À c
p
i
À
Á
¼ n
and
min m p ; m n
À
Á ¼ m
where
m p ¼
X e
i¼1
max c
s
i À c
p
i
À
Á ; 0
À
Á
m n ¼
X e
i¼1
min c
s
i À c
p
i
À
Á
; 0
À
Á
Here c
s
i , c
p
i and e have the same meaning as defined in the case of strong
matching while m p is the sum of vertex surplus and m n is the sum of vertex deficit
in the source distance distribution over the present distance distribution.
The procedure of cycle introduction, canonicalization and unique SMILES
notation generation is the same as done before.
Now, once the structures are generated using the methods described above, one
can use some user-defined parameters incorporated in the computer program to
restrict the number and size of the cycles to be created in the 2D structures. Few
other user-defined parameters, available in the program, may also be used to add
multiplicity of bonds (double and triple bonds) between pairs of vertices and other
hetero-atoms (e.g. nitrogen, oxygen, halogens) in order to get complete 2D structures of the compounds. The output of the generated structures may be saved in
SMILES notations and can be viewed using a molecular modelling software that is
capable of getting molecular structures from SMILES notation. Subsequently, the
activities of the generated structures may be predicted using the rule-based method
[18, 19] standardized for a biological endpoint of interest and can be prioritized and
screened from their MPS values. In this way, one may be able to screen some
potential bioactive compounds from the bigger set of combinatorially generated
molecular structures using topological distance information associated with
activity-related vertices present in the active compounds of a data set under consideration. It may be worth noting at this point that this newly developed method
[15] is essentially a molecular topology-based approach and activity prediction is
done using molecular graphs of the compounds where bond multiplicity and atom
types are not required. However, since bond multiplicity and atom types can be
introduced in the combinatorially generated topological structures using the options
available in the program and those structures can be saved in SMILES format, one
can always use these generated structures for any 2D and 3D drug design/discovery
applications.
Combinatorial Drug Discovery from Activity-Related Substructure …
89