Human protein complex interfaces reveal candidate assembly roles for initiator methionine excision: corrected data and reproducibility archive

This corrected data and reproducibility archive supports the manuscript “Human protein complex interfaces reveal candidate assembly roles for initiator methionine excision.” It asks which human protein N-terminal positions engage assembly partners and whether their interface environments nominate specific roles for methionine removal. The archive combines a human structural census with terminal-boundary sensitivity analyses, evidence-qualified comparisons and partner-resolved proteasome geometry. Its central biological message is that the possible importance of methionine removal depends not only on sequence, but also on which partner an N-terminal position encounters in a particular assembly context. Sequence-class associations and an evidence-qualified atlas identify candidates for direct tests of verified terminal processing, partner engagement, complex assembly and function. The census comprises 7,246 human biological assemblies, 22,291 chain-level observations and 1,191 proteins. The primary sequence-class analysis includes 1,123 analyzable proteins. At the descriptive threshold of at least 25% burial, interface engagement occurs in 37.5% of MetAP-permissive and 47.5% of nonpermissive residue-2 sequence classes; the assembly-chain-count-adjusted odds ratio is 0.65 (p = 7.6 × 10⁻⁴). First-observed coordinate-boundary filtering retains 845 proteins, including 175 interface-engaged permissive-class candidates. These are distinct analysis populations and should not be interchanged with the original 261-protein permissive-class candidate set. The corrected analyses distinguish annotation-selected sequence positions, observed coordinate boundaries, modeled residue identities and independent proteoform evidence. Partner-resolved comparisons cover 22 selected PSMA4, PSMA6 and PSMA7 chain observations across seven PDB entries. They distinguish PSMA7 partner environments, a PSMA6 chaperone-facing backbone and PSMA4 multi-partner packing. Reconstructed group-level tests use explicit membership definitions, eligible backgrounds, boundary sensitivities and multiplicity corrections. A histone-symbol association is supported within the tested structural collection, including peptide-recognition contexts; proteasome enrichment is sensitive to boundary eligibility and multiplicity treatment and is not presented as a robust, universal association. The release provides the corrected supplementary datasets, analysis code, frozen inputs, partner-contact tables, figure source data and artwork, reconstruction outputs, and documentation needed to distinguish current results from labeled historical provenance. The Supplementary Information, main figure legends and five main and three supplementary figures are synchronized with the approved scientific materials of 25 September 2026. The current plotting component contains the exact source inputs and reproduces the approved figure presentation; earlier renderings remain labeled provenance. The original census measurements are retained where appropriate, while corrected labels, sensitivity analyses and revised interpretations are explicitly documented. The synchronization does not add new structural calculations, statistical findings or experimental measurements and is not a census-wide reanalysis. Structural observations are not independent biological replicates, and deposited contexts do not establish an ordered maturation trajectory. Modeled Met1 does not demonstrate physiological methionine retention; first-observed Ser2 does not establish sample-specific excision, acetylation or a free terminal amino group. Incomplete modeled atoms are flagged rather than rebuilt. Pair-only burial values are nonadditive. The archive supports hypothesis generation and reproducible structural analysis, not demonstrated processing causality, disease mechanisms or clinical prediction. This version supersedes release 1.3.0 for reproducing the corrected manuscript analyses. Earlier releases remain historical records and should not be treated as corrected proteoform catalogues. The earlier release is available through its version-specific Zenodo DOI. Cite this corrected archive as version 1.4.0 using DOI 10.5281/zenodo.22966146 when reproducing its results. The associated earlier manuscript, “A structural census links penultimate-residue class to N-terminal burial in human protein assemblies,” is available on bioRxiv, DOI 10.64898/2026.08.31.748389. This archive contains corrections and extensions made after that preprint version; depositing this archive does not update the preprint itself. Author-provided analysis software and associated software documentation are distributed under the MIT License. Author-created derived datasets, figures, the supplementary scientific document and main figure legends are licensed under Creative Commons Attribution 4.0 International (CC BY 4.0), except where third-party rights or notices apply. Copyright © 2026 Yie-Hwa Chang. These are component-specific licenses, not alternative licenses for every file. Third-party inputs retain their applicable terms and attribution requirements. Neither license changes the license of the linked preprint. See LICENSE_SCOPE.md and THIRD_PARTY_NOTICES.md in the archive.