|
|
|
|
|
ABSTRACT:
Motivation
In the present are many QSAR/QSPR models, based on varied considerations, from mathematical through topological and geometrical to 3D molecular geometry approaches.
Idea
The idea is to create a unitary approach, based on a minimal set of well-known truths, capable to generate an efficient model of property behavior depending on molecular structure.
Method
First step in order to reach the proposed goal is to create a huge family of molecular descriptors starting from molecular structure as a graph, considering the bonds and bond types, atom types and a most probable 3D geometry of the molecule. More, using this family of molecular descriptors, a preliminary selection is done in simple linear regression with the measured property. The resulted set of valid descriptors serves for multivariate regressions in order to reach the best QSAR/QSPR model.
Results
The comparisons of the obtained results with other models shows that the proposed model of Molecular Descriptors Family is superior to most of the all other models.
Advantages
The model is dependent only of the microscopic molecular structure and it can be applies at any macroscopic molecular property.
For a given molecular structure or set of structures, is necessary only one calculation of the descriptors, and can be applies to more than one measured property without changes. In other words, the MDF of a molecular structure is a molecular invariant.
Disadvantages
Because the set of molecular descriptors are huge (787968 computed values), the processing time of the model finding is time consuming.
Conclusion
Considering the obtained results, advantages and disadvantages and also the trend of computing performances, the MDF method promise a great expansion of using.
|
|
|
|
STATISTICS
|
|
Click on # to view
|
|
Citations
|
|
16
|
|
References
|
|
0
|
|
Comments
|
|
0
|
|
Quality
|
|
0/0.00
|
|
Interest
|
|
0/0.00
|
|
View(er)s
|
|
1/775
|
|
|
|
|
|
|
| Prev |
Next |
|
