The open-source TRamWAy project developed in our lab aims at inferring the nature and parameters of biomolecule random walks inside living cells based on single-molecule trajectories. One of the approaches available is the mapping of these properties in space. These maps provide quantitative information about biological processes, such as synaptic plasticity and memory development, or HIV particles formation in human T cells. To produce the maps, we assign the points of the molecule trajectories to spatial bins and then calculate the physical parameters in each bin. There are however multiple ways of dividing the region of interest into bins, and one is often faced with a trade-off between getting more data in each bin (larger bins) and getting a higher spatial resolution (smaller bins).
We have recently developed a statistical method, which allows one to quantify information available in individual bins for identification of biological interactions. This statistical test (a Bayes factor) can be further used to evaluate a given spatial mesh and suggest which bins need to be grown or reduced (Fig.). The goal of the proposed internship project is to create an iterative meshing algorithm based on the Bayes factor. Two alternative strategies need to be evaluated: adjusting an existing mesh or constructing a new mesh from scratch by adding extra constraints to standard algorithms. A successfully built meshing algorithm will then be applied to biological data sets, with a potential of providing new insights into analyzed biological process at a fixed minimal level of evidence (Fig.). To the end of the project, the intern will gain experience in Python programming for a leading-edge scientific open-source software project and valuable experience with numerical data analysis algorithms for biological data sets.
A successful candidate should be pursuing a Master’s degree in computer science, physics, or applied mathematics and have programming experience. Prior experience with Python programming, Bayesian statistics, or working with biological or other scientific data will be an advantage.
Further information: Alexander SEROV (alexander.serov __at__ pasteur.fr).