Understanding flavor development during the processing of fine-flavor cocoa using omics technologies and artificial intelligence techniques

The global chocolate market has experienced significant growth over the past decade, with projections indicating it will reach a value of USD 200 billion by 2028. Chocolate production is a complex process, heavily reliant on post-harvesting procedures, including cocoa bean fermentation, drying, and roasting, which profoundly affect the flavor profile of chocolate. The need for standardized and improved cocoa processing methods is currently a pressing challenge to boost the production of high-quality cocoa worldwide as current production relies on artisanal processing methodologies that vary between farms even in the same regions. In this regard, the general objective of this thesis was to understand flavor development during the processing of fine-flavor cocoa using omics technologies and artificial intelligence techniques. Recent studies have delved into cocoa processing through omics analysis, generating a substantial volume of data related to chocolates flavor quality. Consequently, the first specific objective of this thesis (Chapter 1) consisted of a systematic analysis of the existing cocoa omics data and explores opportunities and gaps in the standardization of cocoa processing using data mining techniques. Based on the gaps observed during this systematic review and the fact that metabolomics is the omics data type most easily linkable to the flavor profile, this thesis focuses on a metabolomics analysis of the major cocoa processing steps (fermentation, drying, and roasting). First, fermentation was analyzed separately because it is the most complex step in chocolate production. This is a highly variable and challenging process to control, making it essential to gain insights into the metabolic alterations taking place during this stage to improve the consistency and quality of chocolate production. Thus, the second specific objective of this thesis (Chapter 2) was to understand the changes in the metabolic fingerprint of spontaneous fine-flavor cocoa beans fermentation by MS-based metabolomics. The third specific objective of this thesis (Chapter 3) aimed to dig deeper into cocoa processing biochemistry. The primary objective of this study was to investigate how these key stages of cocoa processing impact the metabolites and lipids linked to the flavor characteristics found in cocoa liquors. This knowledge is crucial for enhancing existing post-harvesting techniques, benefiting both small and large cocoa producers. The third specific objective of this thesis assess the impact of drying and roasting of fine-flavor cocoa beans on the flavor profile using MS-based metabolomics and lipidomic. 11 Finally, the last chapter (Chapter 4) focuses on bridging the flavor profile of cocoa processing with the biochemical changes observed in every stage. The main challenge observed in the previous objectives was the significant number of compounds identified without any association with the flavor profile (less than 40%). Therefore, the last specific objective of this thesis was to unravel the links between the metabolic fingerprint and the flavor profile of fine-flavor cocoa during processing using Machine Learning algorithms. This alternative was selected considering that existing techniques for connecting individual molecules to flavor are both costly and time intensive.

• Autores: Herrera Rocha, Fabio Esteban
• Tipo de recurso: Doctoral thesis
• Fecha de publicación: 2023
• Institución: Universidad de los Andes
• Repositorio: Séneca: repositorio Uniandes
• Idioma: eng
• OAI Identifier: oai:repositorio.uniandes.edu.co:1992/73142
• Acceso en línea: https://hdl.handle.net/1992/73142
• Palabra clave: Machine Learning; Cocoa; Flavor Chemistry; Metabolomics; Ingeniería
• Rights: openAccess
• License: https://repositorio.uniandes.edu.co/static/pdf/aceptacion_uso_es.pdf