Verified Gene-Model Reference Set for Multi-Host Flavor-Trait Engineering: Dataset Descriptor for 28 Pinned Reference Genes and 7,878 Host Orthologs Across 87 Species
We describe the schema, record format, field definitions, and example usage of the Scoville Splice verified gene-model dataset. The dataset comprises four tables: a 28-row reference-pin table anchoring each flavor-trait module to a curated UniProt entry; a 7,878-row host-model table supplying species-specific ortholog sequences, locus coordinates, and quality flags for 87 host species; a module-coverage summary table that maps each host to its resolved and unresolved module slots; and a second-method flag table recording protein-language-model embedding agreement for each host model. Each record is self-auditing: acceptance gates, identity, catalytic-site conservation, copy-number flags, and inferred-terminus flags are exposed as fields rather than hidden inside a composite quality label. This paper is a companion to the methods paper (Ford, 2026b) and is intended for practitioners reading the data records directly. Part of the Intelligent Palate Architecture Platform (IPAP) research series. All records are computational. The methods and constructs described herein are the subject of U.S. Provisional Patent Application Nos. 64/172,827 (filed October 9, 2026), 64/188,197 (filed October 30, 2026), 64/192,062 (filed November 5, 2026), and 64/199,105 (filed November 15, 2026).
Authors
- Priyanca Ford (ORCID: https://orcid.org/0000-0003-0395-1752)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-14
- DOI
- https://doi.org/10.5281/zenodo.24224687
- Primary Topic
- Genomics and Phylogenetic Studies
- Type
- article
- Field-Weighted Citation Impact
- 0.00