Building quantitative prediction models for tissue residue of two explosives compounds in earthworms from microarray...

Area of Science:

• Environmental Science
• Toxicology
• Computational Biology

Background:

• Soil contamination from munitions poses environmental risks, leading to chemical residue in animals.
• Predicting tissue residue is complex due to dynamic biological processes like uptake and transformation.
• High-dimensional gene expression data offers new avenues for computational residue prediction.

Purpose of the Study:

• To develop computational models for predicting tissue residue of 2,4,6-trinitrotoluene (TNT) and 1,3,5-trinitro-1,3,5-triazacyclohexane (RDX) in earthworms.
• To identify key genes associated with tissue residue formation using gene expression data.
• To evaluate the performance of LASSO and Random Forest (RF) algorithms in residue prediction.

Main Methods:

• Analysis of a 240-sample dataset including gene expression and tissue residue of TNT and RDX in Eisenia fetida.
• Application of LASSO and RF computational approaches for predictor gene identification and model building.
• Integration of Wilcoxon rank-sum test and HOPACH for prior variable selection.

Main Results:

• LASSO model demonstrated optimal performance with minimal complexity for TNT residue prediction.
• The combined Wilcoxon-HOPACH-RF approach yielded the highest prediction accuracy for RDX residue.
• Two distinct sets of approximately 30 predictor genes were identified for RDX and TNT, respectively.

Conclusions:

• Both LASSO and RF are effective tools for quantitative prediction of chemical tissue residue.
• The identified gene sets provide insights into the molecular mechanisms of residue formation.
• Further research is needed to refine models and address complex contamination scenarios.