Toward a unified scoring function for native state discrimination and drug-binding pocket recognition

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology
• Structural Biology

Background:

• Protein folding and receptor-ligand recognition are crucial biological processes governed by molecular interactions.
• Current empirical scoring functions are specialized, addressing either protein folding or ligand binding, but not both.
• A unified approach is needed to address complex phenomena like induced fit and intrinsically disordered proteins.

Purpose of the Study:

• To introduce a novel statistical potential for unified protein folding and ligand recognition.
• To evaluate the performance of this new potential against specialized scoring functions.
• To explore its potential applications in induced fit modeling and intrinsically disordered protein docking.

Main Methods:

• Developed a statistical potential based on moieties as fundamental units.
• Applied the potential to benchmark datasets for protein folding and ligand pocket recognition.
• Compared performance against existing specialized scoring functions.

Main Results:

• The new statistical potential demonstrates comparable performance to specialized functions in both folding and ligand recognition.
• The moiety-based approach effectively handles diverse molecular recognition problems.
• The unified framework shows promise for tackling complex biological systems.

Conclusions:

• A single statistical potential can effectively address both protein folding and receptor-ligand recognition.
• This unified approach offers a powerful tool for studying induced fit and intrinsically disordered proteins.
• The moiety-based scoring function advances computational approaches in structural biology and drug discovery.