Raw values for plasma SOMAScan data of CALERIE participants at baseline and 2 years after caloric restriction

Plasma samples of CALERIE participants (n = 42) were subjected to the SomaScan 7K assay. The initial assay and normalization followed SomaLogic’s pipeline. Briefly, SomaScan assay data are initially normalized using hybridization controls to reduce variation within the run caused by readout steps. Subsequently, median signal normalization is performed across pooled calibrator replicates within the run to address within-run technical variations in the calibrator signals before employing them for scaling calculations. The ratios of the reference value to the median of replicates for each SOMAmer are calculated and decomposed into two components: plate scale – representing the median ratio, and calibration scale – representing the recalculated set of scale factors specific to each SOMAmer reagent. The plate scale adjusts for overall signal intensity differences between runs, while calibration adjusts for reagent-specific assay differences between runs. Median signal normalization is conducted using Adaptive Normalization by Maximum Likelihood (ANML) for plasma. For proteomic expression, we used ANML normalization values provided by SomaLogic. The values were log2‑transformed, and proteins that failed internal SomaScan quality control were removed. To identify proteins differentially expressed under CR, we fitted linear models using the limma R package (version 3.56.2), including donor as a blocking factor for paired testing.