CLOUDe

CLOUDe predicts evolutionary targets of gene deletion events from gene expression and sequence data using statistical and machine learning models.

Key Features:

• Ornstein–Uhlenbeck modeling: Models the evolution of gene expression through an Ornstein–Uhlenbeck process capturing the balance between random drift and stabilizing selection.
• Redundant versus unique classification: Classifies deleted genes as "redundant" or "unique" based on sequence characteristics, expression profiles, and molecular functions.
• Machine learning ensembles: Implements multi-layer neural networks, extreme gradient boosting, random forest, and support vector machines for prediction and inference.
• Neural network performance: Multi-layer neural network architecture demonstrates optimal power and accuracy for classifying genes and estimating evolutionary parameters.
• Evolutionary parameter estimation: Estimates evolutionary parameters governing gene expression with high precision.
• Empirical findings: Applied to empirical datasets such as Drosophila and identifies that deletions predominantly target genes with unique functions enriched for protein deubiquitination.

Scientific Applications:

• Inference of deletion targets: Determining whether genes affected by deletions are functionally redundant or unique.
• Evolutionary dynamics estimation: Estimating parameters of gene expression evolution under an Ornstein–Uhlenbeck framework.
• Functional enrichment analysis: Detecting functional enrichments among deleted genes (for example, protein deubiquitination) in empirical datasets such as Drosophila.

Methodology:

Computational methods explicitly include modeling gene expression evolution with an Ornstein–Uhlenbeck process and applying multi-layer neural networks, extreme gradient boosting, random forest, and support vector machines, with the neural network showing superior classification and parameter-estimation performance.