liveBook · Manning

pIC50

As explained in Build AI Drug Discovery Pipelines

pIC50 is the negative base-ten logarithm of a compound's half-maximal inhibitory concentration expressed in molar units.

pIC50 is the negative base-ten logarithm of IC50 measured in molar units. Because the scale reverses concentration, higher pIC50 means greater drug potency. section 3.2.1

Convert IC50 values by first expressing concentration in molars:

Using pIC50 in a dataset

The hERG blockers dataset contains 587 compounds and 526 non-null pIC50 values. Its pIC50 column describes experimental potency, while the classification task predicts hERG blockade; therefore, pIC50 is not used as a training feature. section 3.2.1

Listing 13.2 Gating on binder probability before converting an affinity value to pIC50
GATE = 0.5
 
def score(aff):
    p_binder = aff["affinity_probability_binary"][0]    #A
    if p_binder < GATE:    #B
        return {"binder": False, "p_binder": p_binder, "pIC50": None}
    log_ic50_um = aff["affinity_pred_value"][0]    #C
    return {"binder": True, "p_binder": p_binder,
            "pIC50": 6.0 - log_ic50_um,    #D
            "IC50_uM": 10.0 ** log_ic50_um}

Convert a model's predicted log10(IC50) in micromolar units to pIC50 with pIC50 = 6.0 − log10(IC50 in µM), but only after passing a binder-probability gate. section 13.2.2

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Explains logarithmic conversion with unit examples and thresholds, then stops short of detailed assay methodology or calibration.
Examples
Uses nanomolar and micromolar conversions, a 40 µM hERG threshold, and Python conversion of predicted micromolar affinity.
What it assumes
Requires understanding IC50 and molar concentration while defining both from zero through inhibitory-concentration examples.
What you can do after
Apply via Python to convert IC50 or predicted micromolar log-affinity values and rank compounds by potency.
How it connects
Connects pIC50 to IC50, drug potency, hERG blockers, affinity prediction, and binder-probability gating.

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.