Accurate quantification of circulating proteins is critical for assessing biological variation and integrating proteomics with other omics to understand biological processes and disease mechanisms. Protein measurements, however, can be substantially influenced by preanalytical variability arising from differences in sample collection, handling, and storage, whereas technical variation introduced by the assay and workflow is typically well controlled through established validation procedures. Identifying proteins that capture these systematic influences enables their incorporation into downstream analyzes, thereby improving statistical power. In this study, we applied highly multiplexed aptamer-based affinity proteomics to plasma samples from three independent cohorts─German, Arab-Asian and Qatari to evaluate how adjusting for all measured proteins influences protein quantitative trait loci (pQTLs) associations. Using the p-gain statistic as an indicator of improved association strength, we identified clusters of proteins whose covariation patterns suggested potential preanalytical effects. One cluster contained HSP90 (Heat Shock Protein 90), a marker linked to white blood cell lysis, while others were enriched for proteins involved in complement and coagulation cascades or platelet activation. Our work presents a data-driven framework for detecting latent sources of variation in large-scale proteomic data sets and lay the groundwork for future efforts to quantify the impact of hidden confounding factors.
Stephan et al. (Mon,) studied this question.