TopicNet

TopicNet analyzes and quantifies changes in transcriptional regulatory networks across cellular states by applying latent Dirichlet allocation to gene sets regulated by transcription factors to extract functional topics that capture network rewiring.

Key Features:

• Latent Dirichlet Allocation (LDA) for Functional Topics: Applies LDA to gene sets regulated by specific transcription factors to extract functional topics as low-dimensional representations of regulatory programs.
• Rewiring Score: Computes a rewiring score based on changes in topic composition associated with transcription factors to quantify alterations in regulatory connectivity across states.
• Topic Activity Score: Integrates gene expression data to compute topic activity scores that measure activation levels of functional topics in particular cellular states.
• Differential Survival Analysis: Correlates differences in topic activity with differential survival outcomes in cancers to identify prognostic associations.

Scientific Applications:

• Regulatory Network Analysis: Analyzes large-scale changes in regulatory network connectivity using chromatin immunoprecipitation sequencing (ChIP-seq) data across diverse transcription factors and cell types.
• Oncogenesis Studies: Identifies transcription factors with significant connectivity changes to investigate molecular mechanisms of cancer development and progression.
• Gene Expression Integration: Links regulatory changes to functional outcomes by integrating gene expression to interpret topic activity across cellular states.
• Survival and Prognostic Marker Discovery: Associates topic activity differences with survival outcomes in cancers to support prognostic marker identification.

Methodology:

Applies latent Dirichlet allocation to TF-regulated gene sets to derive topics, computes rewiring scores from topic alterations, integrates gene expression to calculate topic activity scores, performs differential survival analysis on topic activities, and is implemented in R.