Rapid protein domain assignment from amino acid sequence using predicted secondary structure.

Area of Science:

• Structural biology
• Bioinformatics
• Computational biology

Background:

• Determining protein domain content without structural information or sequence homology is a significant challenge.
• Accurate domain identification is vital for understanding protein function and evolution.

Purpose of the Study:

• To evaluate the effectiveness of domain delineation methods in the absence of sequence homology.
• To introduce and assess a novel method, DomSSEA, for continuous domain identification.

Main Methods:

• Comparison of baseline methods, Domain Guess by Size, and the newly developed DomSSEA algorithm.
• DomSSEA utilizes the alignment of predicted secondary structures against observed secondary structures from the CATH domain database.
• Sensitivity was measured by the number of correctly assigned top-scoring predictions.

Main Results:

• DomSSEA achieved a 73.3% success rate in correctly assigning domain numbers to a representative chain set.
• For multidomain proteins, DomSSEA correctly predicted domain number and boundary locations within a +/-20 residue margin for 24% of cases.
• Performance was contextualized against other assessed prediction methods.

Conclusions:

• The developed DomSSEA method offers a viable approach for automatic domain assignment, particularly when sequence homology is limited.
• Alignment of predicted and observed secondary structures provides a valuable basis for identifying continuous protein domains.
• Further refinement may improve accuracy for complex multidomain proteins.