Computational prediction of molecular pathogen-host interactions based on dual transcriptome data.

Area of Science:

• Computational biology
• Systems biology
• Bioinformatics

Background:

• Predicting pathogen-host interactions (PHIs) is crucial for understanding infectious diseases.
• Gene expression data analysis is a key computational method for inferring these interactions.
• Existing methods may not fully capture the complexities of PHIs, such as dynamic environmental changes, response delays, and missing data.

Purpose of the Study:

• To extend the NetGenerator tool for more accurate inference of inter-species gene regulatory networks (GRNs) in PHIs.
• To address specific challenges in PHI data, including complex gene expression patterns, time delays, and missing data points.
• To incorporate gene- and time point-specific variances into the network inference process.

Main Methods:

• Extended NetGenerator, a network inference tool, to model PHI characteristics.
• Tested multiple modeling scenarios, including multi-stimuli perturbation.
• Evaluated the impact of missing data on inference performance.
• Incorporated gene- and time point-specific variances into the objective function and robustness testing.

Main Results:

• Modeling PHI network perturbation with multiple stimuli improves biological representation.
• PHI network inference is feasible even with missing gene expression data, though complete data is recommended.
• Incorporating variance information enhances the robustness of inferred interactions.
• The enhanced NetGenerator successfully predicted previously verified PHIs using dual RNA-sequencing data.

Conclusions:

• The extended NetGenerator provides a more robust computational approach for inferring gene regulatory networks in PHIs.
• The tool effectively handles complex biological scenarios, including missing data and high variance.
• This advancement aids in predicting pathogen-host interactions and understanding disease mechanisms.