Design foresight and model humility for omics inference

Omics inference increasingly proceeds transcriptome-first: measure RNA, then ask how much biology can be extracted later. The ordering should be settled before the assay is chosen, not reconstructed from the data afterward. The first question is what the target requires to be directly observed; the second, what claims the data can honestly support. Design foresight means choosing assays, sampling, and orthogonal measurements based on the question and saying whether transcriptomics is primary, supporting, or insufficient for that claim. Model humility means refusing to promote computational output beyond what the data and assumptions can bear and writing measured, inferred, and predicted results in voices matched to their support. Using transcriptomics as the central case, we argue for question-first omics inference.