Combining Suspect Screening with Large Language Model-Based Text Mining to Comprehensively Characterize Organic...

Area of Science:

• Environmental Health
• Toxicology
• Biochemistry

Background:

• Chemical exposures are linked to maternal pregnancy complications, including gestational hypertension (GH), anemia, and gestational diabetes mellitus (GDM).
• Current research is fragmented, limiting mechanistic understanding due to insufficient compound analysis.
• A comprehensive approach is needed to identify chemical exposures and their roles in pregnancy complications.

Purpose of the Study:

• To develop a novel framework integrating high-throughput analysis and large language model text mining.
• To identify organic compounds associated with pregnancy complications and create an exposure atlas.
• To elucidate the specific chemical signatures and pathogenic mechanisms underlying GH, anemia, and GDM.

Main Methods:

• Integrated high-throughput analysis with large language model-based text mining.
• Identified novel organic compounds in human milk and those not previously linked to pregnancy complications.
• Utilized text mining to uncover disease-specific chemical signatures and pathogenic pathways.

Main Results:

• Identified five novel compounds in human milk and 35 additional compounds linked to pregnancy complications.
• Established disease-specific chemical signatures: GH (PAHs), anemia (nitrogen compounds), GDM (carboxylic acids).
• Revealed pathogenic mechanisms: oxidative stress (GH), red blood cell damage (anemia), mitochondrial dysfunction (GDM).

Conclusions:

• Constructed an atlas of organic compounds associated with pregnancy complications.
• Provided new insights into the environmental origins and mechanisms of these complications.
• The novel framework enables comprehensive understanding and identification of chemical exposures impacting maternal health.