It is impossible to think of a sustainable world without the effective participation of microorganisms in the development of bioprocesses. The research of doctoral student Maurício Alexander Moura Ferreira, from the Graduate Program in Agricultural Microbiology at UFV, focuses on the development of optimization tools for predicting the use of proteins in metabolic models. Two of these tools, which can be used to rationalize the application of microorganisms in the development of sustainable industrial processes, have just been presented in two recently published articles.

The first is “Parrot”, a tool that compares the abundance of proteins between an already known model and a second situation in which the cell is exposed to stress. “The trick is that with Parrot, we can see how the enzymes are being used in this second model, compared to the first case. Through these comparisons, we understand how the use of the enzymes differs, which can guide us and increase the efficiency of our bench work,” explains Mauricio.

The text presenting the tool, “PARROT: Prediction of enzyme abundances using protein-constrained metabolic models”, signed by Maurício, his advisor, Professor Wendel Silveira, and his co-supervisor, Zoran Nikoloski, from the University of Potsdam in Germany, was recently published in the journal PLOS Computational Biology. The main contribution of Parrot, in Maurício’s words, is an improvement in the way models are applied to metabolic engineering. “It improves because it compares two situations – between the ideal, which is already known, and the new thing. This allows us to optimize the testing stage, because it makes it more precise and allows a direct comparison between two situations. And an interesting thing is that these different situations that we test are precisely conditions that exist in industry.”

The model was developed with data relating to bacteria and yeasts, but it is possible to use it with any living being, including humans. “The application with human beings would be interesting, we have many metabolic models today that test data from the health area, so it’s totally applicable.”

Artificial intelligence
The second tool, Camel, also uses metabolic models to calculate the use of enzymes, but adds artificial intelligence to fill a gap left by Parrot. “One issue with Parrot is that we can only calculate a certain amount of proteins, while the cell produces a bit more. This is a mathematical limitation. So we developed a tool that combines metabolic modeling and artificial intelligence to be able to make this leap, to encompass this ‘little bit more’ that Parrot doesn’t catch,” explains Maurício.

Again, the tool was developed with data related to bacteria and yeasts, but it will be available for use in other areas. “The most immediate application may be for other scientists working in metabolic engineering, precisely to improve the production of biomolecules, such as biofuels, ingredients for the food industry, pharmaceuticals, etc., and to use this to guide experimental work on the bench. And the idea is that, with this, we can have more efficient, faster and more precise metabolic engineering and genetic engineering strategies.”

The second article, “Accurate prediction of in vivo protein abundances by coupling constraint-based modelling and machine learning”, published in the prestigious Metabolic Engineering journal of the International Metabolic Engineering Society, adds to the authors the names of Philipp Wendering and Marius Arend, linked to Professor Nikoloski’s laboratory, where Maurício was during his sandwich period.

Both papers are part of Maurício’s doctoral research. He is currently working on a third computational biology tool, and it is the application of all three that will lead to the thesis he will present to the PPGMBA soon.