Modeling sequence and function similarity between proteins for protein functional annotation.

Area of Science:

• Bioinformatics
• Computational Biology
• Molecular Biology

Background:

• Protein function prediction is crucial in biological research.
• Pair-wise sequence alignment, like BLAST, is commonly used but relies on a simplistic homology concept.
• The relationship between protein sequence similarity and functional similarity is complex and not always linear.

Purpose of the Study:

• To model the relationship between protein sequence and function similarity.
• To gain insights into the sequence-function similarity relationship for improved protein function prediction.
• To evaluate the effectiveness of current methods and identify areas for improvement.

Main Methods:

• Development of a model to quantify sequence and function similarity between proteins.
• Analysis of the variability in function similarity across different levels of sequence similarity.
• Assessment of the limitations of existing sequence alignment algorithms.

Main Results:

• Function similarity generally increases with sequence similarity, but with significant variability.
• High sequence similarity is often required to ensure high functional similarity.
• Current multiple sequence alignment algorithms face challenges with low-similarity sequences.

Conclusions:

• Simple sequence similarity thresholds are insufficient for accurate protein function prediction.
• Profile models offer higher sensitivity but require robust multiple sequence alignment methods.
• Further development of multiple sequence alignment algorithms is needed to handle diverse sequence similarities effectively for protein function prediction.