ATEN

ATEN infers Boolean network topology and dynamics from short, noisy gene expression time series by using an ensemble of And/Or trees to identify predictive prime implicants.

Key Features:

• And/Or Tree Ensemble Algorithm: Uses an ensemble of And/Or trees per target gene to identify prime implicants, i.e., conjunctions of input genes predictive of the target gene state.
• Boolean Network Inference: Infers Boolean functions for each target gene from selected implicants and assembles these functions into a Boolean network representing regulatory interactions.
• Handling Noisy Data: Designed to infer networks from short and noisy gene expression time series typical of experimental datasets.
• Reproducibility in Parallel Computing: Provides clusterSetRNGStream() to manage random number streams in parallel environments, addressing limitations of set.seed().
• Parameter Tuning: Enables adjustment of the number of And/Or trees (parameter B), with larger B recommended for larger or noisier datasets at the cost of increased computation.

Scientific Applications:

• Systems Biology: Reconstructs regulatory network topology and dynamics to study gene interactions and regulatory mechanisms governing cellular processes.
• Experimental Time-Series Analysis: Infers regulatory interactions from short, noisy gene expression time series produced by experimental studies.
• Artificial and Real-World GRN Studies: Applies to both synthetic and empirical gene regulatory network datasets for modeling, validation, and hypothesis generation.

Methodology:

For each target gene, ATEN selects prime implicants using an ensemble of And/Or trees, uses those implicants to infer Boolean functions that predict gene state transitions, and combines the inferred functions into a Boolean network.