On the Unfounded Enthusiasm for Soft Selective Sweeps III: The Supervised Machine Learning Algorithm That Isn’t

In the last 15 years or so, soft selective sweep mechanisms have been catapulted from a curiosity of little evolutionary importance to ubiquitous mechanism claimed explain most adaptive evolution and, in some cases, evolution. This transformation was aided by series articles Daniel Schrider and Andrew Kern. Within this series, paper entitled “Soft sweeps are dominant mode adaptation human genome” (Schrider Kern, Mol. Biol. Evolut. 2017, 34(8), 1863–1877) attracted great deal attention, particular conjunction with another (Kern Hahn, 2018, 35(6), 1366–1371), for purporting discredit Neutral Theory Molecular Evolution (Kimura 1968). Here, we address an alleged novelty Kern’s paper, i.e., claim that their study involved artificial intelligence technique called supervised machine learning (SML). SML is predicated upon existence training dataset which correspondence between input output known empirically be true. Curiously, Kern did not possess genomic segments priori evolved either neutrally through hard sweeps. Thus, using thoroughly utterly misleading. absence legitimate datasets, used: (1) simulations employ many manipulatable variables (2) system data cherry-picking rivaling worst excesses literature. These two factors, addition lack negative controls irreproducibility results due incomplete methodological detail, lead us conclude all inferences derived so-called algorithms (e.g., S/HIC) should taken huge shovel salt.