Data and code: representation of WHO-reported human outbreaks of eight priority zoonotic pathogens in public sequence archives, and documented human sequencing for the same events, 2020-2025

Retrieved data, derived results and the complete analysis code for a cross-sectional audit asking whether human outbreaks of WHO priority zoonotic pathogens reported through WHO Disease Outbreak News are represented by a human genome in the public sequence archives, and, for the same events, whether published evidence of human sequencing exists at all. Unit of analysis: a pathogen-country-year for which at least one Disease Outbreak News item published between 1 January 2020 and 31 December 2025 named that country in its title. Updates on one outbreak that straddled a calendar-year boundary (last item of year Y and first of year Y+1 no more than 75 days apart) are counted as one event, dated Y. Outcome: at least one nucleotide record of that pathogen with host Homo sapiens, that country and that collection year in the INSDC index served by NCBI Virus, with records created after 27 October 2026 excluded. NCBI Virus can index a record some time after it is created, so this cut-off does not freeze the index: a re-retrieval on 4 November 2026 with the same cut-off returned 253 additional records and changed none of the 52 event outcomes. Pathogens: MERS-CoV, Nipah virus, Lassa virus, Crimean-Congo haemorrhagic fever virus, ebolaviruses, Marburg virus, monkeypox virus and Rift Valley fever virus. Contents: cross_pathogen_audit.py runs the retrieval and analysis end to end with no credentials and prints every number reported in the associated manuscript, including four sensitivity analyses and a positive control on events reported in 2013-2019. raw_taxa/ holds every nucleotide record of each taxon as indexed by NCBI Virus, raw_don/ the complete Disease Outbreak News index (3201 items), and results_cross_pathogen.json the 52 events with their outcomes, the continuation merge, the excluded items and the control events. ratings_per_event.tsv holds the two raters' independent classification of every event in the blinded literature audit, and audit_numbers.py regenerates every number in the manuscript from the result files. Scripts that build the figures, tables, supplement and reporting checklist are included. Third-party raw sequence data are not redistributed; the cached files hold record metadata only. The manuscript text itself is not deposited before publication. Changes from version 2.0.0 (version 2.1.0, 4 November 2026): a companion archive, protocol_and_verification_2026-11-04.zip, adds the pre-registered literature-audit protocol (Arabic original, SHA-256 7500365cd950f0847b62f98286dad9d07c92f026fbe53e24a20d27747ef254a2, with an English translation) and the verification run of 4 November 2026: the records missed by retrieval on familiar pre-binomial taxon names in NCBI Virus and Entrez Nucleotide, the offline reproduction and live re-retrieval of the main analysis, and the script for the table of literature-audit ratings. The main archive is unchanged. Changes from version 1.0.0: species-level NCBI taxon identifiers replace the no-rank child identifiers used before, which had missed records assigned directly to the species after ICTV renaming; the index date is taken from the actual retrieval cut-off rather than being written into the code; one outbreak that had been counted as two events is merged; and the per-event ratings of the blinded two-rater literature audit are added. The headline result changes from 25 of 53 to 26 of 52 events represented.