Automatic discovery of errors in models of metabolism
Stable isotope tracing with modern mass spectrometry is a powerful method that allows tracking the metabolism of nutrients in living human cells and tissues at the atom level. For example, by “labeling” a sugar with carbon-13 isotopes at specific atoms, we may follow its conversion into fat in the liver or in fat tissue, and also monitor activity of the metabolic enzymes involved in this process. This methodology makes heavy use of mathematical models of human metabolism to interpret complex data in a systematic way. However, these models often contain errors such as missing enzymes, which can be difficult and time-consuming to find.
In this project, we want to investigate the use of L1 regression, an established technique for outlier detection, to automatically detect errors in our metabolic models. The project involves performing simulation studies to investigate “in silico” how effective the L1 regression method is at detecting such errors, and performing tests of the method on data from isotope tracing experiments in human cells and tissues. If successful, we hope this method can accelerate our studies of human metabolism and metabolic disease by removing tedious and error-prone manual work. The basic computational methods and frameworks involved are already in place, but some modifications may be necessary.
We seek a M.Sc student with a background in computer science, engineering physics, applied mathematics or equivalent. You should have good experience with programming and be comfortable with technical computing platforms such as Matlab, Mathematica, or Python/numpy. Good understanding of multidimensional analysis and optimization methods are an advantage. Knowledge of cell and molecular biology is not necessary, but some familiarity with chemistry/biochemistry can be helpful.
The project is envisioned as a M.Sc degree project, but might also be performed as an internship. Part of the project work can be done remotely. You will work together with and be advised by experienced researchers in computational biology and experts in metabolism.
For more information, please contact roland.nilsson@ki.se
Computational analysis of metabolism in human fat
Fat cells are central for human health, and dysfunctional fat can lead to metabolic disorders like diabetes and fatty liver disease. In this project, we aim to investigate metabolic changes that occur in adipocytes (fat cells) in development of obesity, using the flux balance analysis technique. In this method, we construct a mathematical model of the biochemical reactions in adipocytes, and fit this model to data from actual human adipocytes to understand how metabolism functions. Practically, this involves solving linear or quadratic programming problems with established methods, implementing changes to the model to explain the observed data, and interpreting the results. The data includes measurements of uptake and release of metabolites from the cells, as well as thermodynamic data on reactions.
We seek a M.Sc student with a background in biotechnology, biochemistry, engineering biology, or equivalent. You should have basic knowledge of biochemistry and in particular metabolism, at the level of basic biochemistry courses; some experience with technical computing platforms and programming languages, such as Matlab, Mathematica, R or Python; and a solid understanding of linear algebra. Knowledge of optimization methods and linear programming in particular is an advantage.
The project is envisioned as a M.Sc degree project, but might also be performed as an internship. Part of the project work can be done remotely. You will work together with and be advised by experienced researchers in computational biology and experts in metabolism.
For more information, please contact roland.nilsson@ki.se