Food Allergy Detection With Ml: Developing Machine Learning Algorithms To Detect And Prevent Allergens In Food Products

Abstract

Synthetic biology and genome editing are two examples of cutting-edge technology that have opened the door to the production of new foods and functional proteins. It is important to appropriately assess their toxicity and allergenicity, nevertheless. Some proteins are known to cause allergies due to certain sequences of amino acids, however the identities of many of these sequences are yet unknown. Here, we present a data-driven strategy and a machine-learning technique for discovering previously unknown allergen-specific patterns (ASPs) in amino acid sequences. The proposed method allows for a thorough search for subsequences of amino acids that are statistically more common in proteins that cause allergies. For this proof-of-concept, we used the proposed technique on a database consisting of 21,154 proteins for which the allergy status is already known. This proof-of-concept investigation found ASPs that were in agreement with previous biological findings. Using these ASPs improved allergenicity prediction performance compared to existing methods, suggesting this approach could be valuable for assessing the practicality of synthetic meals and proteins.