The Four Horsemen of the ‘Omicsalypse’: ontology, replicability, probability and epistemology

Much of modern genomics and the other ‘omics’ that tag along, assert that the causal bases of biomedical outcomes are genomically enumerable lists whose effects are predictable with ‘precision’, extensible from samples to all, and enabled by ever-greater hypothesis-free data accumulation. The assertion rests on fundamental, if often implicit assumptions, that (1) the phenomena are based on underlying law-like biological causation, and, therefore, are (2) replicable and (3) even if not deterministic, have specifiable, stable, essentially parametric, probabilities, all of which (4) essentially equates induction with deduction, enabling asymptotically accurate prediction based on past observation. These glowing promises are the four horsemen of a genocentric ‘Omicsalypse’. But what if the assumptions are wrong or appropriate only to an extent that is unknowable, even in principle? Might there be better ways to understand complex traits?