Cassandra – Quality modeling using detailed genomics

Many dairy starter cultures consist of complex microbial communities of closely related strains where subtle genetic differences often result in distinctive phenotypic features that define product characteristics including flavour. Diacetyl, acetoin, acetate, acetaldehyde and ethanol are central compounds for the quality and distinct flavor profile of many dairy products. These compounds are produced during complex fermentation processes. The aim of this project is to develop a tool that can predict formation of specific desired and undesired fermentation compounds using only the information about microbial community structure, fermentation conditions and dairy substrate. 

By: Grith Mortensen

Many dairy products (e.g., buttermilk and many cheese types) are produced using undefined starter cultures containing several different species further divided into dozens of different strains. Understanding such processes are far from trivial, as important technological properties differ at the bacterial strain level and interactions play a strong role in determining the final quality. The researchers will in this project use long read DNA sequencing technologies, which give new opportunities to study complex microbial cultures with precision never seen before. Detailed genomics combined with high-resolution data on aroma formation during fermentations open new possibilities for developing machine learning-based tools. Tools that enable rapid prediction of desired and unwanted formation of compounds in the final fermented product via a single sequencing process.

The objective of Cassandra is thus to establish a system for the prediction of specific compounds in dairy products based on bacterial strain-level characterization using long-read sequencing, bioinformatics, and data analysis. The final output will be an easy-to-use workflow covering both laboratory and data analysis steps. The new in-silico modeling will save time, resources and money in dairy product development as well as surveillance of fermentation processes by replacing the traditional time consuming, tedious and expensive “trial and error” methods.