TISCalling

TISCalling leverages machine learning to identify and rank translational initiation sites across eukaryotes, with particular emphasis on non-canonical TISs in plants and viruses.

Key Features:

• Machine Learning-Based Prediction: Combines machine learning models and statistical analysis to identify and rank novel translational initiation sites.
• Cross-Kingdom Generalization: Generalizes and ranks important features common to multiple plant and mammalian species while identifying kingdom-specific features such as mRNA secondary structures and "G"-nucleotide contents.
• Viral TIS Prediction: Achieves high predictive power for identifying novel viral translational initiation sites.
• Prediction Score Visualization: Provides prediction scores for putative translational initiation sites along plant transcripts to enable prioritization for further validation.
• Sequence-Aware Interpretability: Offers an interpretable approach for decoding genome sequences and exploring functional proteins.

Scientific Applications:

• Novel Protein and Peptide Discovery: Identifies genes encoding novel proteins or small peptides through recognition of translational initiation sites.
• Non-Canonical TIS Identification: Enables systematic and global identification of non-AUG translational initiation sites in plants that conventional computational methods cannot address.
• Feature Importance Evaluation: Evaluates the importance of mRNA sequence features for translational initiation site determination.
• Plant Transcript Prioritization: Prioritizes translational initiation sites of interest for experimental validation in plant transcripts.
• Viral Genome Analysis: Identifies novel translational initiation sites in viral genomes.

Methodology:

TISCalling combines machine learning models and statistical analysis to identify and rank novel translational initiation sites.