Assembling multidomain protein structures through analogous global structural alignments.

Area of Science:

• Structural biology
• Computational biology
• Protein structure prediction

Background:

• Proteins often function cooperatively through multiple domains.
• Current methods primarily focus on single-domain structures, creating a gap in multidomain modeling.
• High demand exists for accurate multidomain protein structure models.

Purpose of the Study:

• To develop an efficient pipeline for constructing multidomain protein structures.
• To address the limitations of existing methods in modeling proteins with multiple domains.
• To enable automated, genome-scale assembly of multidomain protein structures.

Main Methods:

• Developed the DEMO (Domain Assembly) pipeline for multidomain structure construction.
• Utilized docking-based domain assembly simulations.
• Determined interdomain orientations using distance profiles from analogous templates via domain-level structure alignments.
• Incorporated sparse restraints from mass spectrometry cross-linking data and cryo-EM density maps.

Main Results:

• DEMO achieved correct global fold (TM-score > 0.5) in 86% of continuous and 100% of discontinuous domain cases on a benchmark set.
• Applied to CASP12 and CASP13, DEMO models showed significantly improved quality over original server predictions.
• Integration of experimental data (cross-linking, cryo-EM) improved TM-score by 6.3% and 12.5%, respectively.

Conclusions:

• DEMO provides an efficient and accurate approach for multidomain protein structure assembly.
• The pipeline is suitable for automated, genome-scale structural modeling.
• DEMO enhances the quality of protein structure models, especially when incorporating experimental data.