A Multistrategy AI‐Assisted Consensus Screening Pipeline for Discovery of Candidate p53 Monomer Binders

Abstract

The rapid advancement of AI has greatly boosted the efficiency of large‐scale lead compound screening, especially for proteins without resolved 3D structures in the PDB. However, current large structure databases cannot support efficient large‐scale screening within drug R&D cycles. Here, we propose a multistrategy consensus screening pipeline and establish a corresponding small‐sample rapid screening system to identify candidate mini binders as potential lead ligands for the tumor suppressor p53 monomer. Using AI models, we predicted the p53 monomer structure and built a 500‐mini‐protein library. Protein–protein docking, co‐folding, and energy scoring were integrated into the consensus screening pipeline, with the PD‐L1&VHH6 complex as a positive control. Through comprehensive computational evaluation via the constructed consensus screening pipeline, A0A0R0G186 was determined as the optimal hit candidate with a calculated interfacial hydrophobic energy density ( γ bind  = −3.47 kcal mol −1  Å −2 ). Structural alignment demonstrates that its key binding region shows high structural similarity to the native protein A0A0R0G186 (RMSD = 0.34 Å). The candidate protein exhibits a predicted amyloid formation probability of 0.787, a large proportion of soluble regions, and a reasonable hydrophobic core distribution, which computationally indicates favorable druggability potential. Notably, all the above findings are computational predictions that require further systematic experimental validation.