NRpred

NRpred classifies nuclear receptors into subfamilies using support vector machines on amino acid and dipeptide composition for sequence-based functional categorization.

Key Features:

• Methodology: Support vector machine (SVM) classifiers utilize amino acid composition and dipeptide composition derived from nuclear receptor sequences.
• Data Source: The classifiers were developed using a non-redundant dataset of 282 nuclear receptor proteins sourced from the NucleaRDB database.
• Validation Approach: Classifier performance was assessed using 5-fold cross-validation.
• Performance Metrics: Amino acid composition-based classifiers achieved 82.6% overall accuracy, while dipeptide composition-based classifiers achieved 97.5% overall accuracy.

Scientific Applications:

• Functional annotation of nuclear receptors: Classification supports analysis of nuclear receptor transcription factors that regulate gene networks controlling cell growth, differentiation, and homeostasis.
• Disease-related investigations: Subfamily-level classification aids studies of receptor misregulation implicated in diseases such as diabetes, osteoporosis, and cancer.

Methodology:

Support vector machine classifiers were trained on amino acid composition and dipeptide composition features derived from a non-redundant set of 282 nuclear receptor proteins from NucleaRDB and evaluated by 5-fold cross-validation.