Add DockQ v2 protein complex interface quality scoring

Description of feature

Summary

This issue proposes adding DockQ v2 as a new interface quality scoring module
This work will be started during the nf-core Hackathon March 2026.

Background

When predicting protein complexes (co-folding), it is not sufficient to evaluate
the overall structural accuracy - the quality of the binding interface between
the two proteins is equally important, as this is where biological function occurs.

DockQ v2 addresses this by providing:

• iRMSD — interface RMSD (accuracy at the contact surface)
• LRMSD — ligand RMSD (accuracy of the smaller chain)
• fnat — fraction of native contacts correctly predicted
• DockQ score — single combined interface quality score (0 to 1)

This makes DockQ v2 a natural complement to US-align (Issue #515):
US-align evaluates global structure, DockQ evaluates the binding interface specifically.

Proposed implementation

• No official Docker image exists — custom Dockerfile required
• Install via pip: pip install DockQ
• Source: https://github.com/bjornwallner/DockQ
• Input: predicted complex PDB + reference complex PDB
• Output: iRMSD, LRMSD, fnat, DockQ score
• New nf-core module: dockq/score

Resources

• DockQ v2 source: https://github.com/bjornwallner/DockQ
• DockQ v2 paper: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1011474

Tasks

• Write custom Dockerfile for DockQ v2
• Push image to GitHub Container Registry or Docker Hub
• Write nf-core module for dockq/score
• Connect as optional evaluation step after co-folding
• Test with predicted vs reference complex PDB pair
• Document usage in pipeline README