Integrating Proteomic Studies on Aging

Aging proteins were identified by integrating the information of four independent studies using the SOMAscan platform for proteomic measurements (Sathyan et al., 2020b (link); Arthur et al., 2021 (link); Ferkingstad et al., 2021 (link); Robbins et al., 2021 (link)). Two studies stated that they used SomaScan version 4 (Sathyan et al., 2020b (link); Ferkingstad et al., 2021 (link)) and two studies reported a ‘5k assay’ (Arthur et al., 2021 (link); Robbins et al., 2021 (link)). As several SOMAmers target similar proteins this provides a resolution on proteoform level, however for readability we refer to ‘proteins’ across this study and an overview of the number of included aptamers across studies is presented in Table 1. Two studies provided raw SOMAscan data of the plasma proteome (Arthur et al., 2021 (link); Robbins et al., 2021 (link)), while the other two provided summary statistics of their analyses for all measured plasma proteins (Sathyan et al., 2020b (link); Ferkingstad et al., 2021 (link)). For a full overview of the included studies and demographics, see Table 1. For the studies providing raw data, we performed linear modeling to test for the effect of age on protein expression levels, while correcting for most of the available metadata to correct for possible confounding effects. Proteins were defined as significantly associated to aging at a Benjamini-Hochberg (FDR) adjusted p-value (q) below 0.01 (q < .01).