Predictivity performances as squared error (SE) on the validation ( n = 27 samples, d ) datasets for Stabl L (left), Stabl EN (middle) and Stabl AL (right). Stabl SRM performances are shown using MX knockoffs. Results using random permutations are shown for Stabl L in Supplementary Table 5 . Median and IQR values comparing Stabl SRM performances to their cognate SRMs are listed in Supplementary Table 5 .
