Scholarlink

Computational protein modeling is a cornerstone of modern biomedical engineering, linking molecular insights with drug discovery. We present a project-based pipeline that guides students through sequence retrieval, structure prediction using four methods (SWISS-MODEL, MODELLER, RoseTTAFold, AlphaFold3), refinement via RosettaRelax, and validation with MolProbity and RMSD superposition in ChimeraX. Applying this workflow to human and eel Na⁺/Cl⁻ cotransporter (NCC), we achieved high structural convergence-monomer backbone RMSDs below 2 Å and dimer RMSDs below 3 Å-while identifying method-specific features such as SWISS-MODEL’s secondary interfacial cavity in the dimer. CavityPlus analysis pinpointed both conserved central pockets (the known thiazide-binding site) and novel cavities suitable for ligand binding. Key findings include AlphaFold3’s unparalleled geometric accuracy (over 98% of residues in favored Ramachandran regions) and faithful replication of the drug-binding pocket. This integrated approach not only clarifies the strengths and differences of each software tool but also provides a robust foundation for future membrane protein simulations and targeted drug-design efforts.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Comparative Pipeline for Computational Protein Modeling: A Guide for Biomedical Engineering Students

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Julio Rafael Sandria Sánchez,
  • Gerardo Rodríguez Cárdenas,
  • Cristian Borislavov Gotchev Chtereva,
  • Erika Moreno,
  • Luis Jimenez-Angeles

摘要

Computational protein modeling is a cornerstone of modern biomedical engineering, linking molecular insights with drug discovery. We present a project-based pipeline that guides students through sequence retrieval, structure prediction using four methods (SWISS-MODEL, MODELLER, RoseTTAFold, AlphaFold3), refinement via RosettaRelax, and validation with MolProbity and RMSD superposition in ChimeraX. Applying this workflow to human and eel Na⁺/Cl⁻ cotransporter (NCC), we achieved high structural convergence-monomer backbone RMSDs below 2 Å and dimer RMSDs below 3 Å-while identifying method-specific features such as SWISS-MODEL’s secondary interfacial cavity in the dimer. CavityPlus analysis pinpointed both conserved central pockets (the known thiazide-binding site) and novel cavities suitable for ligand binding. Key findings include AlphaFold3’s unparalleled geometric accuracy (over 98% of residues in favored Ramachandran regions) and faithful replication of the drug-binding pocket. This integrated approach not only clarifies the strengths and differences of each software tool but also provides a robust foundation for future membrane protein simulations and targeted drug-design efforts.