To delay or prevent Alzheimer’s disease (AD) we must understand the molecular mechanisms underlying how and why it begins. AD develops over decades but our molecular understanding of the disease from humans is built almost entirely on analyses of deceased brains where the initiating stresses may be absent or masked by later-stage pathological processes and compensatory responses. The currently favoured transgenic rodent models (based on ectopic expression and/or multiple mutations in genes causing early onset familial AD, EOfAD) have been selected because they mimic somewhat the histopathology thought to define AD. However, the most detailed form of molecular analysis currently available (transcriptome analysis) shows these models bear little relationship to the human disease (Hargis & Blalock, 2017, Behav Brain Res 322:311). We posit that, in the absence of an AD hypothesis with reliable predictive power, the most objective approach to genetic modelling (making fewest assumptions) is to copy the genetic state of EOfAD as closely as possible – i.e. heterozygosity for a single mutation in an EOfAD gene. Mouse knock-in models were generated 15+ years ago but never analysed using modern ‘omics technologies. Six years ago the Alzheimer’s Disease Genetics Laboratory began a program using genome editing technologies to introduce EOfAD-like (and other) mutations into zebrafish. By comparing transcriptomes and proteomes in young adult (pre-pathology) brains from different mutants we aim to identify the “signature” molecular changes-in-common that define the stresses eventually leading to AD. Currently, we have isolated mutations in the zebrafish orthologues of the EOfAD genes PSEN1, PSEN2, APP, and SORL1. ‘omics and other analyses are underway on heterozygous mutants and their wild type siblings at various ages. Unexpected molecular phenomena have been observed and the first outlines of a molecular signature of EOfAD may be discerned. Our first comparative analyses support that cellular energy stress initiates EOfAD.