1. Identify the descriptor with the highest correlation to activities (D1a), i.e. the highest R2 value.
2. Rank remaining descriptors according to their correlation with D1a.
3. Create a table with row and column of descriptors arranged in the order of their correlation with D1a.
4. Tabulate the Rsq values of QSAR equations consist of exactly two descriptors.
Common trends:
1. Descriptors with low correlation and negative correlation with D1a usually are not useful in raising the goodness-of-fit of the QSAR equation.
2. Descriptors around certain correlation with D1a tend to yield QSAR equation with higher Rsq.

No comments:
Post a Comment