DomainMapper: Accurate domain structure annotation including those with non-contiguous topologies.

Area of Science:

• Structural bioinformatics
• Computational biology
• Protein domain analysis

Background:

• Automated domain annotation is crucial for structural informatics.
• Current methods using hidden Markov model (HMM) profiles struggle with non-contiguous or insertional domains.
• Ambiguous annotations arise from complex protein topologies.

Purpose of the Study:

• To develop an algorithm, DomainMapper, for accurate protein domain structure annotation.
• To address challenges posed by non-contiguous and insertional domains.
• To provide a robust tool for structural informatics pipelines.

Main Methods:

• Developed the DomainMapper algorithm for unique domain structure assignment.
• Validated annotations using the AlphaFold database.
• Analyzed domain non-contiguity prevalence in yeast and human proteomes.

Main Results:

• DomainMapper accurately annotates protein sequences, including those with complex topologies.
• Non-contiguous domains are prevalent, found in 10.74% of yeast and 4.52% of human domains.
• Identified specific protein folds with a propensity for non-contiguous or insertional structures across the Tree of Life.

Conclusions:

• DomainMapper enhances the accuracy of automated protein domain annotation.
• The algorithm effectively handles proteins with non-contiguous and insertional domains.
• Findings highlight the widespread nature of non-contiguous domains and their evolutionary implications.