Machine Learning and Network Toxicology Analysis of DEHP Targets in Glioma

A research document by Ren Li, last updated in May 2026, integrates network toxicology and machine learning to study the plasticizer DEHP's role in glioma. The study identifies 77 overlapping genes and prioritizes 12 key genes, validated with a diagnostic model achieving a pooled AUC of 0.994. In vitro experiments on U87 cells confirmed dose- and time-dependent upregulation of core targets like RELA, ABCA1, and HIF1A.

Use Cases

• Prioritizing environmental exposure biomarkers for glioma based on the identified 12 key genes.
• Building diagnostic models for glioma using the described machine learning framework and gene signatures.
• Studying protein-ligand interactions through the molecular docking and dynamics simulations of DEHP with core proteins.
• Conducting functional enrichment analysis on pathways like neuroactive ligand-receptor interactions and GABAergic synapses identified from the 77 overlapping genes.

Strengths

• Integrates data from four major databases (CHEMBL, CTD, SwissTargetPrediction, PharmMapper) and two public genomic repositories (GEO, TCGA).
• The resulting 12-gene diagnostic model demonstrated strong external validation with a pooled AUC of 0.994 across multiple cohorts.
• Findings were experimentally validated in U87 glioma cells, showing dose- and time-dependent effects.

Limitations

• The dataset is a 643.1 KB DOCX file, which is a small document format; the underlying raw data tables are not directly provided.
• Column-level documentation and sample data are unavailable, making field semantics and structure unclear.
• The temporal and geographic scope of the source genomic data is not specified, which may limit reproducibility for specific populations.

Provenance

Source

figshare, authored by Ren Li.

Collection Method

Computational analysis integrating network toxicology and 127 machine learning models on public genomic data, followed by in vitro validation.

Freshness

Last updated 2026-05-07 22:02:33