Integrated correlation network analysis for advanced chemical profiling of consumer products: a proof-of-concept for the volatile organic compounds in pulps and paperboards.

The comprehensive chemical profiling of consumer products is essential for ensuring consumer safety, yet increasingly sophisticated datasets generated by modern analytical instrumentation urgently necessitate advanced interpretive strategies. Here we propose a proof-of-concept strategy integrating 3 distinct correlation network analyzes to identify the interactions among variables from a high-dimensional dataset. By utilizing comprehensive two-dimensional gas chromatography - high-resolution mass spectrometry and chemical space interpretation, 628 identified VOCs are mapped to demonstrate the inherent compositional complexity of 21 pulps and paperboards. Unlike conventional chemometrics only provide isolated categorical markers, regularized and debiased sparse partial correlation network analyzes reveal 267 valid pairwise connections encompassing co-varying patterns, carryover and baseline effects, competitive pathways, and distinct chemical signatures across 203 VOCs. Weighted correlation network analysis further discerns 2 highly significant module-trait correlations clustering 31 VOCs with synchronized patterns. The distinct mathematical foundations of these approaches impart high complementarity and synergy for resolving variable connections in high-dimensional datasets, especially those featuring extreme sample-variable imbalance, thus establishing a highly effective and scalable framework for chemical profiling of consumer products characterized by inherent compositional complexity.