pathway parameter advising

pathway parameter advising optimizes parameter selection for pathway reconstruction algorithms by assessing topological similarity between generated and curated pathways using a graphlet decomposition metric.

Key Features:

• Graphlet Decomposition Metric: Uses a graphlet decomposition metric to quantify topological similarity between generated pathways and manually curated pathways from pathway databases.
• Parameter Advising Algorithm: Tunes parameters of pathway reconstruction algorithms in a method-agnostic manner to minimize biologically implausible predictions.
• Evaluation and Performance: Evaluated against other parameter selection methods across four pathway reconstruction algorithms, demonstrating improved avoidance of implausible networks and accurate reconstructions using NetPath data.
• Implementation: Implemented in Python 3.6 using NetworkX and NumPy.

Scientific Applications:

• Pathway Reconstruction from Omic Data: Improves the biological plausibility of reconstructed pathways derived from condition-specific omic data.
• Benchmarking Parameter Selection: Provides a framework for comparing parameter selection strategies across different pathway reconstruction algorithms.
• Influenza Host Factor Network Construction: Applied to construct an influenza host factor network.
• Use of Curated Databases: Leverages curated pathways from the NetPath database as background knowledge to guide parameter selection.

Methodology:

Computes graphlet decomposition metrics to assess topological similarity to curated pathways, ranks or selects parameter sets to minimize implausible network topologies, compares selected parameters against alternatives across four pathway reconstruction algorithms, and is implemented in Python 3.6 using NetworkX and NumPy.