A structural homology approach for computational protein design with flexible backbone

Area of Science:

• Protein engineering
• Computational biology
• Structural bioinformatics

Background:

• Structure-based Computational Protein Design (CPD) is crucial for protein engineering.
• Current CPD methods rely on energy functions, which are imperfect.
• Incorporating known structural data can enhance CPD accuracy.

Purpose of the Study:

• To introduce Shades, a data-driven CPD method.
• To improve sequence design by exploiting local structural environments and energy functions.
• To accommodate mutations by sampling side-chain and backbone conformations.

Main Methods:

• Shades utilizes customized libraries of non-contiguous in-contact amino acid residue motifs.
• The method combines local structural environments from known proteins with energy functions.
• It samples side-chain and backbone conformations to facilitate mutations.

Main Results:

• Shades achieved 30% protein sequence recovery and 46% sequence similarity against PFAM families on a benchmark of 40 proteins (excluding homologs).
• Including homologous structures increased wild-type sequence recovery to 93%.
• The method demonstrated improved performance in protein sequence design.

Conclusions:

• Shades offers a data-driven approach to enhance Computational Protein Design.
• Integrating structural information significantly improves sequence recovery and similarity.
• The method provides a valuable tool for advancing protein engineering and functional protein design.