GNIPLR

GNIPLR infers directed gene regulatory networks (GRNs) from time-series and non-time-series gene expression datasets by applying projection techniques and lagged regression.

Key Features:

• Projection Techniques: GNIPLR employs LASSO projection (LSP) and linear projection (LP) to transform gene expression data into a linear and monotonous pseudo-time series.
• Directional Inference via Lagged Regression: It applies lagged regression analyses to infer the directionality and potential causal relationships between genes.
• Data-Type Compatibility: The method supports both time-series and non-time-series gene expression datasets for constructing directed topological networks.
• Validation and Performance: Validation on simulated and real biological data, including liver hepatocellular carcinoma (LIHC) and bladder urothelial carcinoma (BLCA) expression datasets, showed higher accuracy and Area Under the Curve (AUC) relative to comparative methods.

Scientific Applications:

• Systems Biology: Constructing GRNs to elucidate complex biological processes and regulatory mechanisms.
• Cancer Transcriptomics: Inferring regulatory interactions in cancer expression datasets such as LIHC and BLCA.
• Method Benchmarking: Comparing GRN inference accuracy and AUC using simulated and real biological datasets.

Methodology:

Gene expression data are projected using LASSO projection (LSP) and linear projection (LP) to create a pseudo-time series, followed by lagged regression analysis to determine the directionality of gene regulation.