IRESPred

IRESPred predicts Internal Ribosome Entry Sites (IRES) in viral and cellular mRNA 5' untranslated regions (UTRs) to identify cap-independent translation initiation signals.

Key Features:

• SVM-Based Prediction: Employs a Support Vector Machine (SVM) algorithm to classify IRES versus non-IRES sequences.
• Comprehensive Feature Set: Uses 35 features derived from sequence and structural properties of UTRs, including probabilities of interactions between UTRs and small subunit ribosomal proteins (SSRPs).
• High Predictive Performance: Reported blind-test performance: accuracy 75.51%, sensitivity 75.75%, specificity 75.25%, precision 75.75%, and Matthews Correlation Coefficient (MCC) 0.51.
• Comparative Advantage: Outperforms existing tools such as VIPS, which is limited to viral IRES prediction.

Scientific Applications:

• Viral Research: Identification of viral IRES elements to study viral cap-independent translation and hijacking of host protein synthesis machinery.
• Cellular Biology: Prediction of IRES in cellular mRNAs to investigate alternative translation initiation and regulation of gene expression under physiological and pathological conditions.

Methodology:

Integrates sequence and structural features from UTRs with interaction probabilities between UTRs and SSRPs to generate 35-feature input vectors, which are classified using a Support Vector Machine (SVM).