liveBook · Manning

Correlation

As explained in Build AI Drug Discovery Pipelines

The Pearson correlation coefficient is a statistic ranging from -1 to 1 that measures the strength and direction of a linear association between two continuous variables.

Pearson correlation measures the strength and direction of a linear association between two continuous variables. Its coefficient, r, ranges from -1 to 1: values near -1 indicate a strong negative relationship, values near +1 indicate a strong positive relationship, and 0 indicates no linear relationship. section 4.1.3

In molecular modeling, compare computed molecular descriptors with a target such as solubility. Scatterplots reveal distributions, outliers, and apparent relationships before correlation quantifies them. Increasing logP is associated with decreasing solubility, while hydrogen-bond donors show a slight positive association; TPSA appears uninformative for this target. section 4.1.3

Use correlation to support practical modeling decisions:

  • Inspect feature–target relationships before model fitting. section 4.1.3
  • Identify descriptors that may contribute little linear information to a target and consider them during feature engineering. section 4.1.3
  • Evaluate whether predicted and true values move together; high positive correlation supports correct relative ranking even when predictions have systematic bias. section 11.6.2

Binding-affinity evaluation

For DGraphDTA, Pearson correlations of 0.867 on Davis and 0.903 on KIBA indicate strong positive linear relationships between predicted and true affinities. These values complement error metrics such as MSE by showing how well the model preserves relative differences. section 11.6.2

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Defines the coefficient, interprets its range, and applies it to descriptors and model predictions without deriving the statistic.
Examples
Uses logP, hydrogen-bond donors, TPSA, solubility, DGraphDTA predictions, and Davis and KIBA affinity datasets.
What it assumes
Assumes continuous variables and basic familiarity with molecular descriptors, solubility, or binding-affinity predictions.
What you can do after
Evaluate linear associations and model ranking quality with Pearson correlation alongside error metrics.
How it connects
Connects Pearson correlation with molecular descriptors, solubility, feature engineering, binding affinity, MSE, and model evaluation.

Also explained in

Compare how 22 books explain correlation →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Build AI Drug Discovery Pipelines. The section links point to the parts of the book it draws on.