이전 사이트 이동

Plant-Derived Allergenic Protein Sequences

A comprehensive plant protein dataset was constructed for allergenicity prediction by integrating plant-derived allergen sequences and non-allergen protein sequences from publicly available allergen databases and UniProt. The allergen dataset includes proteins from plant-related sources such as pollen, seeds, nuts, fruits, vegetables, cereals, legumes, and other plant taxa. Non-allergen proteins were collected from UniProt and assigned to the plant category based on organism-level taxonomy information. Where possible, non-allergen proteins were selected from the same or related plant-source organisms as those represented in the allergen dataset to reduce organism-level bias. This design helps the model focus on allergen-associated sequence features rather than simply distinguishing proteins by biological source. In total, 13,186 plant-derived protein sequences were compiled, including 1,927 allergen sequences and 11,259 non-allergen sequences, providing a curated resource for protein language model-based allergenicity prediction and explainable deep learning analysis.