neurosnap.algos.pdockq module#

neurosnap.algos.pdockq.calculate_pDockQ(structure, chain1=None, chain2=None, dist_thresh=8.0)[source]#

Calculate the predicted DockQ (pDockQ) score and corresponding PPV for a two-chain complex.

Parameters:

structure (Union[str, Protein]) – Protein object or a path/ID string that can be parsed by neurosnap.protein.Protein
chain1 (str) – Chain ID for the first chain (e.g., “A”). If None, auto-detected if only 2 chains exist.
chain2 (str) – Chain ID for the second chain (e.g., “B”). If None, auto-detected if only 2 chains exist.
dist_thresh (float) – Distance threshold (Å) for interface contacts, default 8.0

Returns:

pdockq (float): Predicted DockQ score between 0 and 1. Higher values suggest greater confidence that the predicted interface is correct.
ppv (float): Estimated positive predictive value (precision) corresponding to the pDockQ score. Represents the minimum expected probability that the predicted interface is correct (bounded roughly 0.55-0.98 with higher being better).

Return type:

(pdockq, ppv)