Type 2 diabetes (T2D) is a major cause of morbidity, mortality, and high healthcare costs worldwide. Diet has been identified as an important modifiable risk factor of T2D. However, the biological mechanisms underlying pathophysiology of T2D and how diet might affect molecular processes that may be involved in T2D remain largely unclear. In SMCC and COSM cohorts, we aim to (1) identify protein biomarkers that could predict incident T2D; (2) investigate whether predictive proteins may provide new insights to the pathways related to disease development and may be used for improving risk prediction beyond established risk factors; (3) investigate the impact of diet in modifying proteins associated with T2D risk. To achieve these aims, multivariate-adjusted cox proportional hazards regression will be used to estimate hazard ratios of T2D risk for proteins. We will apply machine learning based variable selection techniques to identify most relevant protein predictors and evaluate their incremental value in risk prediction compared to established risk factors. Dietary patterns will be derived using a data-driven approach and thereafter, associations of proteins with food items and/or dietary patterns will be assessed using Partial Spearman correlation analysis, accounted for covariates. The present study constitutes the hitherto largest study where a high number of proteomic biomarkers targeted for cardio-metabolic risk factors are assessed from a prospective cohort to find protein biomarkers of T2D and their relation to diet. This will provide excellent possibilities to discover novel biomarkers of T2D, as well as provide potential targets for nutritional prevention and therapy for T2D, with large potential implications for public health.