Fractal Custody Objects: route-comparable chain-of-custody for deterministic computational biology and AI-agent provenance

The contribution targets two well-documented computational-biology problems: the reproducibility crisis in computational biology (Papin etal. 2020; Ziemann et al. 2023) and biomedical data-integrity / tamper-evidence in thedata-to-figure chain (Bik et al. 2016), now compounded by AI-assisted analysis that can assert unverifiable identifiers. The method is demonstrated on a real public mass-spectrometry file (ProteoWizard mzML) and motivated by intrinsic-disorder prediction reproducibility (CAID). It is a provenance/custody method for computational biology — not a generic computer-science result — and the worked example, citations, and framing are biological/biomedical throughout.