BioAutoML

BioAutoML automates machine learning workflows for biological sequence analysis by performing feature extraction, feature selection, algorithm recommendation, and hyperparameter optimization.

Key Features:

• Automated Feature Engineering: Extracts numerical representations of biological sequences using the MathFeature package and performs automated feature extraction and selection.
• Metalearning-Based Model Optimization: Recommends machine learning algorithms and performs hyperparameter tuning using Automated Machine Learning (AutoML) strategies.
• Integrated Machine Learning Pipeline: Supports end-to-end processing from sequence feature generation to optimized predictive model development.

Scientific Applications:

• Noncoding RNA Classification: Predicts major classes of noncoding RNAs (ncRNAs) using machine learning models trained on sequence-derived features.
• Bacterial ncRNA Prediction: Identifies eight categories of bacterial ncRNAs, including housekeeping and regulatory RNAs.
• Biological Sequence-Based Prediction: Enables machine learning analyses of biological sequence datasets through automated feature generation and model optimization.

Methodology:

BioAutoML extracts numerical features from biological sequences using MathFeature, performs automated feature selection, and applies metalearning modules for algorithm recommendation and hyperparameter tuning to optimize machine learning models.