Related Experiment Videos
Enhanced automated function prediction using distantly related sequences and contextual association by PFP.
Troy Hawkins1, Stanislav Luban, Daisuke Kihara
1Department of Biological Sciences, College of Sciences, Purdue University, West Lafayette, Indiana 47907, USA.
Summary
Protein Function Prediction (PFP) improves automated protein function annotation by analyzing weakly similar sequences. This method enhances accuracy and coverage, aiding interpretation of vast experimental data.
Area of Science:
- Bioinformatics
- Computational Biology
- Genomics
Background:
- Exponential growth in experimental data overwhelms protein annotation.
- Lack of reliable annotations hinders interpretation for uncharacterized proteins.
- Existing methods struggle with proteins lacking experimental characterization or homologs.
Purpose of the Study:
- Introduce PFP, an automated function prediction server.
- Enhance protein function annotation coverage and accuracy.
- Provide probable annotations across Gene Ontology (GO) branches.
Main Methods:
- Extend PSI-BLAST search with individual GO term extraction and scoring.
- Utilize a Function Association Matrix for scoring annotation pairs.
- Lower prediction resolution for less predictable functions to increase coverage.
Main Results:
- PFP significantly improves accuracy and coverage (>5-fold) over standard PSI-BLAST.
- Accurate function assignment using weakly similar sequences.
- Predicted annotations (GO depth >= 8) achieve >60% accuracy and ~100% coverage.
Conclusions:
- PFP offers a robust solution for automated protein function prediction.
- The method effectively annotates proteins with limited homology.
- PFP demonstrates superior performance in benchmark assessments.