Improving automatic GO annotation with semantic similarity

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Automatic protein functional annotation is a significant challenge in bioinformatics.
• Manual annotation is time-consuming and resource-intensive.
• There is a need for computational tools to automate protein annotation.

Purpose of the Study:

• To extend the GrAPFI method for automatic protein annotation with Gene Ontology (GO) terms, creating GrAPFI-GO.
• To incorporate semantic similarity and hierarchical relations of GO terms into the annotation process.
• To evaluate the performance of the proposed method.

Main Methods:

• Adapted the graph-based automatic protein function inference (GrAPFI) method.
• Developed GrAPFI-GO for protein annotation using GO terms.
• Explored similarity measures based on common neighbors in protein similarity graphs.
• Implemented a pruning and post-processing technique considering GO term semantic similarity and hierarchy.

Main Results:

• The GrAPFI-GO method was compared with and without common neighbor similarity.
• Performance of GrAPFI-GO and other annotation tools was tested.
• Experimental results demonstrated the effectiveness of the proposed approach.

Conclusions:

• The proposed semantic hierarchical post-processing improves GrAPFI-GO performance.
• The method also enhances the performance of other annotation tools.
• GrAPFI-GO offers an efficient procedure to improve automatic protein function annotation by exploiting GO term semantic relations.