Digital Salmon – from a reactive to a pre-emptive research strategy in aquaculture
New! Highlights 2020
Our research in 2020 focused on completing a high-quality metabolic model of Atlantic salmon. Our model captures how enzymatic reactions convert nutrients in the fish feed to the fillet that we eat, and it allows us to connect the Atlantic salmon genome to growth and feed utilisation. It performs very well in standardised tests and captures expected metabolic (in)capabilities of salmon such as amino acid requirements for growth. We are now using modelling to make predictions for aquaculture such as growth-limiting amino acids in current commercial feeds and possible compositions for novel feeds based on sustainable plant-based ingredients.
Besides completing our salmon metabolic model, a major highlight of 2020 was receiving funding for a DigiSal spin-off project where we are extending our modelling efforts to the salmon gut microbiota. In collaboration with other groups at NMBU and the University of Minnesota, we are modelling the spatiotemporal dynamics of microbes in the salmon gut to understand why a specific bacterium becomes dominant when juvenile salmon transition from fresh to salt water.
Being part of a transdisciplinary centre provides us with connections to the systems biology community and industry and helps us see things from a different perspective. It also incentivises collaboration, within projects as well as between different projects (e.g. workshops with the dCod project). Meeting scientists from highly diverse backgrounds, often in informal settings, has been very valuable for us, and it has allowed us to see opportunities for working together with others that we would not have discovered otherwise.
Scientific publications 2020: 3
Yang Jin, Rolf Erik Olsen, Thomas Nelson Harvey, Mari-Ann Østensen, Keshuai Li, Nina Santi, Olav Vadstein, Atle M. Bones, Jon Olav Vik, Simen Rød Sandve, Yngvar Olsen
Jesse van Dam, Jasper J Koehorst, Jon Olav Vik, Vitor A.P. Martins dos Santos, Peter J. Schaap, Maria Suarez-Diez
Alex Kojo Datsomor, Rolf Erik Olsen, Nikola Zic, Angelico Madaro, Atle M. Bones, Rolf Brudvik Edvardsen, Anna Wargelius, Per Winge
Zdenka Bartosova, Susana Villa Gonzalez, Per Bruheim
Zdenka Bartosova, Marit Hallvardsdotter Stafsnes, Andre Voigt, Per Bruheim
Teshome Dagne Mulugeta, Torfinn Nome, Thu-Hien To, Manu Kumar Gundappa, Daniel J. Macqueen, Dag Inge Våge, Simen Rød Sandve, Torgeir Rhoden Hvidsten
Thomas Nelson Harvey, Simen Rød Sandve, Yang Jin, Jon Olav Vik, Jacob Seilø Torgersen
Alex Datsomor, Nikola Zic, Keshuai Li, Rolf Erik Olsen, Yang Jin, Jon Olav Vik, Rolf Edvardsen, Fabian Grammes, Anna Wargelius, Per Winge
Yang Jin, Inga Leena Angell, Simen Rød Sandve, Lars-Gustav Snipen, Yngvar Olsen, Knut Rudi
Yang Jin, Rolf Erik Olsen, Mari-Ann Østensen, Gareth Benjamin Gillard, Keshuai Li, Thomas Nelson Harvey, Nina Santi, Olav Vadstein, Jon Olav Vik, Simen Rød Sandve, Yngvar Olsen
Discovering connections between salmon genes and fish feed
Norwegian researchers are developing a mathematical model of the salmon that will make it easier to find the optimal feed for aquaculture production of salmon.
The salmon is a predator which in the wild eats small fish and crustaceans in the ocean, and in the first decenniums of salmon breeding the feeding where based on fish oil and fish meal. But the growth in the business has demanded other feed sources, and today most of the feed are produced on land by farming. 75 percent of the fat and protein in the feed are derived from different plants and crops. This is not a sustainable feed source, and also the prices varies a lot.
The business is therefore investigating alternative feed variants, but to test out which of the many alternatives that are the best choice for salmon is time consuming, with a potential cost of several millions. An additional challenge is that the best feed mixture may vary considerably between different salmon strains.
The researcher know a lot about the salmon genes as the whole salmon genome was fully sequenced and mapped in 2016. They also know a lot about salmon physiology from studying the intestine, liver and muscles. Currently, the aim is to develop a mathematical model for all the processes going on in the fish's different organs, and then fuse it into one coherent model of salmon physiology - the digital salmon. This will then be used to simulate and test different variations of salmon feed, thereby allowing faster and more cost efficient development of better and more sustainable feeds.
Researchers within mathematics, data analysis, informatics, sensor technology, genomics and experimental biologists work together in the project.
The project is headed by the Norwegian University of Life Sciences (NMBU) with partners at NTNU, UiB, UiT,University of Stirling, UK, University of Wagening (The Netherlands), and the Institute of Marine Research (Norway).
Industrial partners are AquaGen (salmon breeding) and EWOS (feed producer).
More about the project
Towards the Digital Salmon
In the project Towards the Digital Salmon: From a reactive to a pre-emptive research strategy in aquaculture (DigiSal) the researchers will establish a mathematical model of the salmon physiology, to aid the development of tomorrow's salmon feed. The long term vision is to construct the Digital Salmon - a system physiology model where all the salmon's body functions are simulated in a coherent mathematical model. This will be the core in an open access knowledge platform of salmon.
Todays feed composition used for salmon farming is not sustainable. New ingredients are tried out, such as yeast and bacterial extracts, and micro algae. It is, however, a time consuming and expensive process to test out all possible combinations of alternative ingredients in feeding experiments. Also, different salmon strains will most likely respond differently due genetic differences. By replacing initial experiments by simulations, a more cost effective and improved feed development can be established.
Utilizing the salmon genome
The researcher will therefore translate knowledge about salmon physiology into mathematical expressions. The underlying biochemical pathways are tightly linked to the genetic expression of the enzymes and regulatory proteins involved. A major achievement was made in 2016 when the complete salmon genome was released after sequencing three billions of DNA bases making up the salmon's 29 chromosomes. Mapping the genome also revealed how salmon strains and the salmonidae family may have evolved.
This new insight paved the way for efficient measurement of the level of expression of all genes in a tissue or blood sample from salmon. In addition, measurements of thousands of biomolecules and metabolites in the same samples can be fed into the model. Based on the data, a dynamic model that respond to variations in feeding, genetic background and other parameters will be constructed.
The knowledgebase in terms of the data and the model, will enable new analysis of already known aspects of the salmon's biology, discover knowledge gaps, further acquiring and incorporating new data, and there by develop and improve the model. The goal is that eventually this will be a tool used by the aquaculture industry to test out variants of salmon strains, feed and environment in a simulation, thereby foreseeing eventual challenges and problems before they occur in production. Hence, the title "from reactive to pre-emptive".
Mathematical models make it easier
The researcher view the salmon as a biological system built up of a set of components that rely on each other. To understand variation in a given phenomenon, such as body growth, the potential reactions and processes in the different organs contributing to the phenomenon are combined in a mathematical model that fits with the experimental measurements that can be sampled from salmon.
The benefit of describing physiological processes by mathematical models and run simulations is partly that the researcher can reveal new knowledge, and partly that wrong hypothesis' can be abandon before being tested in experiments. The researcher can therefore quickly select the feed mixtures that most likely will succeed, and then test these in actual feeding experiments.
In DigiSal the researcher especially study metabolism and how this is coupled to genes. They can measure what the salmon eats, how the feed is ingested and converted down to molecular level.
The mathematical model will be encoded with standardized nomenclature of the salmon's biochemical reactions, enzymes, reaction parameters, and molecules involved. The standardization allows automatic coupling of already established biochemical databases and datasets form other experiments.
The project gathers researchers that do mathematics, data analysis, informatics, sensor technology, genomics and experimental biology.
As so many disciplines and partners are involved, the project management emphasizes that the different experts should develop understanding of each others research areas and encourages the team to see possibilities and limitations within the project.
Most of the researchers are located at NMBU, Ås, but several of the project partners are located in other parts of Norway or abroad. The project partners will gather physically from time to time, but usually meetings and contact will be by Skype.
Responsible Research and Innovation
The project will emphasis the dialogue with lay people, researchers within and outside the project, industrial partners and representatives for the authorities. During the project seminars aimed at lay people discussing the use of systems biology for more sustainable food production, possibilities for industry development, and potential impacts of new technology on the consumer, will be arranged.
The project will also make data and models available and usable by a web-platform, integrated with a resource base of the salmon genome.
When the mathematical models are in place, the fish farmers will get recommendations about the feed mixture of their specific salmon strain. In the future, the fish farmers might get custom advice on dosing of the feed, incorporating weather and temperature data and forecasts, for optimal feeding.
For companies that develop new feed variants, the mathematical models will be a tool to select the best ingredients, resulting in shorter time from research to market with less failing animal experiments and lower costs.
For companies that breed new salmon strains, the model will enable them to speed up the process of selecting suitable fishes for further development and breeding, as well as quality control, surveillance and maintenance of their product,
Following the project and an increased need for monitoring and data acquisition, one could envisage development of devices and sensors for a future marked in aquaculture.