for structure generation, both of them lead to the generation of a topological
structure of another active compound. The details of the strong active range are
given in Table 3 and the structure generation details in Fig. 8.
The compound no. 13 along with its molecular graph and the chosen structure
generation vertex (root vertex) is given in Fig. 7a. The distance distribution associated with this vertex (Vertex No. 17) starting with distance 0 is (1, 2, 2, 1, 1, 1, 1,
3, 5, 1, 1). A sample rooted tree is shown in Fig. 7b with the corresponding distance
distribution.
Considering any rooted tree, cycles can be introduced (described in the methods
section) to generate the topology of the structural formula of variety of chemical
compound while still maintaining the distance distribution. In the present study, we
have chosen to generate structures containing two cycles, having number of sides 5
or 6, to investigate whether we are able to generate any other active compound
present in the studied data set. A number of structures are generated in the process,
and it has been found that the structures generated from the root vertex of compound no. 13 contain one such structure that matches with that of compound no.
19 (Fig. 8). It is interesting to note that compound no. 19 is an active compound
from the test set (Table 2) which shows that the method can generate a structure
that it has not seen in the training set. Therefore, one can expect to design novel
structures using this method.
3.1.2 Studies with Nucleoside Analogues
For the nucleoside analogues (NA), we have carried out activity prediction and
structure generation studies. It may be noted that for this series of compounds, we
have investigated the performance of the training set–test set identification tool
using the corresponding algorithm incorporated in the computer program. As
mentioned earlier, in this way we are able to obtain a suitable training set for the
system’s learning and predict activities of the compounds on the basis of this
Fig. 8 One of the structures generated, from compound no. 13, which resembles the topology of
compound no. 19 (Table 1)
94
Md.I. H. Rizvi et al.
structure of another active compound. The details of the strong active range are
given in Table 3 and the structure generation details in Fig. 8.
The compound no. 13 along with its molecular graph and the chosen structure
generation vertex (root vertex) is given in Fig. 7a. The distance distribution associated with this vertex (Vertex No. 17) starting with distance 0 is (1, 2, 2, 1, 1, 1, 1,
3, 5, 1, 1). A sample rooted tree is shown in Fig. 7b with the corresponding distance
distribution.
Considering any rooted tree, cycles can be introduced (described in the methods
section) to generate the topology of the structural formula of variety of chemical
compound while still maintaining the distance distribution. In the present study, we
have chosen to generate structures containing two cycles, having number of sides 5
or 6, to investigate whether we are able to generate any other active compound
present in the studied data set. A number of structures are generated in the process,
and it has been found that the structures generated from the root vertex of compound no. 13 contain one such structure that matches with that of compound no.
19 (Fig. 8). It is interesting to note that compound no. 19 is an active compound
from the test set (Table 2) which shows that the method can generate a structure
that it has not seen in the training set. Therefore, one can expect to design novel
structures using this method.
3.1.2 Studies with Nucleoside Analogues
For the nucleoside analogues (NA), we have carried out activity prediction and
structure generation studies. It may be noted that for this series of compounds, we
have investigated the performance of the training set–test set identification tool
using the corresponding algorithm incorporated in the computer program. As
mentioned earlier, in this way we are able to obtain a suitable training set for the
system’s learning and predict activities of the compounds on the basis of this
Fig. 8 One of the structures generated, from compound no. 13, which resembles the topology of
compound no. 19 (Table 1)
94
Md.I. H. Rizvi et al.