Celiac disease T-cell epitope reference (Sollid 2020 nomenclature), machine-readable

A machine-readable transcription of Table 1 of Sollid et al. (2020), the canonical nomenclature and listing of celiac disease-relevant gluten epitopes recognised by CD4+ T cells. The source table cannot be extracted by text parsers: its rows are vector-outlined in the published PDF, and part of its meaning is carried typographically. Bold marks glutamate formed by transglutaminase 2 (TG2) deamidation that is important for T-cell recognition; underline marks additional glutamine also targeted by TG2. This dataset converts that typography into explicit data fields. Contents: 38 named epitope entries over 31 distinct 9-residue sequences, across HLA-DQ2.5, DQ2.2, DQ8 and DQ8.5 restrictions. Each row carries the binding register, deamidation positions, additional TG2 target positions, unmarked (native) glutamate positions, source protein, grain, previous names, reference, provenance tier, and links to other entries sharing the same sequence. Two counts are reported deliberately and answer different questions: 38 named entries, 31 distinct sequences. Six sequences appear under more than one name, both across HLA restrictions and across grains. Four entries carry five glutamate residues that are NOT deamidation sites. These are recorded explicitly, because treating every glutamate as a deamidation site is the most likely error a consumer of this data would make. Method: transcribed twice independently from 600-dpi renders and machine-diffed across every transcribed field, with the build aborting on any unresolved disagreement. Disagreements require an adjudicated record giving both readings and the basis for resolution. Independently verified by a third reader who re-read all 38 rows directly from the source PDF at 900-1600 dpi: zero discrepancies in sequences, deamidation positions, TG2 target positions or markers. Produced with AI assistance, directed and reviewed by the depositor; see the README for the disclosure and for the development error log. The scientific content is entirely that of Sollid et al. (2020). Please cite that paper. This dataset adds machine-readability, verification and derived fields, nothing more. This is a research reference, not health information. The presence of a peptide here is not evidence that any food is unsafe, and no dietary decision should be based on it.