SFM: A novel sequence-based fusion method for disease genes identification and prioritization

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Identifying disease genes is crucial for advancing disease diagnosis and treatment.
• Existing machine learning methods for disease gene identification face challenges with prior knowledge utilization, negative data selection, and classification strategies.

Purpose of the Study:

• To propose a novel Sequence-based fusion method (SFM) for identifying disease genes.
• To overcome limitations of existing methods by utilizing universal amino acid sequence data instead of noisy prior knowledge.

Main Methods:

• Developed SFM using four distinct feature vectors derived from amino acid sequences.
• Implemented a distance-based approach to generate four negative sets and selected the intersection for more reliable negative data.
• Employed a Decision Tree (C4.5) as a fusion method to integrate results from four independent Support Vector Machine (SVM) predictors.

Main Results:

• The proposed SFM achieved a precision of 82.6%, recall of 85.6%, and F-measure of 84%.
• Experimental evaluations confirmed the method's effectiveness and validity compared to existing approaches.

Conclusions:

• The Sequence-based fusion method (SFM) demonstrates significant efficiency and accuracy in identifying disease genes.
• Utilizing amino acid sequences and a robust fusion strategy offers a promising approach for disease gene discovery.