liveBook · Manning

Dissociation constant

As explained in Build AI Drug Discovery Pipelines

A dissociation constant is an equilibrium measure of the tendency of a molecular complex to dissociate, with lower values indicating stronger binding.

The dissociation constant (Kd or KD) measures how tightly a ligand binds its target. Lower values mean stronger binding. For kinetic measurements, KD equals koff divided by kon and is reported in molar units. section 6.1.3 section 11.1.1

Use Kd as an experimentally grounded target in a drug-discovery workflow:

  • Measure binding with surface plasmon resonance (SPR): inject ligand over immobilized target, derive kon during association, and derive koff during dissociation. section 6.1.3
  • Convert raw Kd values reported in nanomolar units to pKd, using pKd = -log10(Kd), so higher values represent stronger affinity and large ranges become easier to model. section 11.1.1 section 11.6.2
  • Treat assay limits carefully: the DAVIS value of pKd 5 aggregates interactions measured at 10,000 nM or greater, rather than representing one common biological affinity. section 11.1.1
  • Evaluate predicted pKd with error and ranking metrics: MSE measures numerical accuracy, while Pearson correlation and concordance support relative-affinity prioritization. section 11.6.2

Interpreting a prediction error

An RMSE of 0.45 pKd units corresponds to about a 2.8-fold typical error in Kd, because one pKd unit represents a tenfold change in dissociation constant. Such predictions can distinguish high-affinity binders from weak binders and help rank compounds for virtual screening. section 11.6.2

Figure 6.4 Surface plasmon resonance (SPR) plot of response signal to time, and how changes in the angle of the reflected light correspond to the underlying biomolecule binding events that are taking place at the surface layer, between the immobilized target and the flow of solution containing the ligand of interest.

Inspect the SPR response curve to connect association and dissociation phases with the kinetic quantities used to calculate KD. section 6.1.3

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Explains definition, SPR measurement, logarithmic transformation, assay artifacts, and prediction-error interpretation without deriving binding thermodynamics.
Examples
Uses HIV-1 TAR ligand measurements, DAVIS and KIBA affinity datasets, and DGraphDTA prediction results.
What it assumes
Defines association, dissociation, SPR, Kd, and pKd before applying them to affinity prediction.
What you can do after
Evaluate/understand affinity measurements, transform Kd into pKd, interpret model errors, and prioritize compounds by predicted binding strength.
How it connects
Connects dissociation constant to binding affinity, kon, koff, surface plasmon resonance, pKd, QSAR, graph neural networks, and virtual screening.

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.