Annotation transfer for genomics: measuring functional divergence in multi-domain proteins.

Area of Science:

• Genomics
• Bioinformatics
• Structural Biology

Background:

• Annotation transfer is crucial for genome annotation, relying on sequence similarity to infer function.
• Multi-domain proteins are prevalent in eukaryotes but pose greater challenges for accurate functional annotation transfer.
• Previous studies focused on single-domain proteins, leaving multi-domain proteins under-explored.

Purpose of the Study:

• To conduct a large-scale survey of annotation transfer in multi-domain proteins.
• To assess the impact of shared domain combinations and sequence similarity on functional conservation.
• To compare annotation transfer reliability between single-domain and multi-domain proteins.

Main Methods:

• Utilized SCOP superfamilies to define protein domain folds.
• Employed a SWISS-PROT keyword thesaurus for functional categorization.
• Analyzed large datasets of protein sequences and their annotations.

Main Results:

• Multi-domain proteins exhibit lower functional conservation than single-domain proteins, except when sharing identical domain fold combinations.
• Annotation transfer certainty for multi-domain proteins sharing one domain fold is 35%, compared to 67% for single-domain proteins.
• Sharing two domain folds increases functional conservation probability to 80% for multi-domain proteins, exceeding 90% with full-length coverage.
• Functional divergence is approximately two-fold greater in multi-domain proteins compared to single-domain proteins at similar sequence similarity levels.

Conclusions:

• The structural complexity of multi-domain proteins inherently reduces the accuracy of annotation transfer.
• Specific combinations of domain folds are key predictors of conserved function in multi-domain proteins.
• Understanding these differences is vital for improving the robustness of genome annotation pipelines.