Algoritmer för simulering av stokastiska modeller i molekylärbiologi

Tidsperiod: 2012-01-01 till 2014-12-31

Projektledare: Per Lötstedt

Finansiär: Vetenskapsrådet

Bidragstyp: Projektbidrag

Budget: 1 862 000 SEK

Computer simulation is a tool for better understanding of the reactions between molecules and their diffusion in living cells. There are now experimental data for single molecules to compare with and for calibration of the parameters in the models. There is an inherent randomness due to the small number of molecules in the cells and only stochastic models can reproduce what is observed in experiments. In a mesoscopic model, the state of the system is given by the number of molecules of each chemical species in voxels of a discretized space. In a microscopic model, single molecules move by Brownian motion and react with each other with a certain probability when they collide. These two levels of modeling will be merged in an efficient and accurate multiscale algorithm. The most accurate and expensive description at the microscale will be used only when it is necessary for the accuracy in the simulation. The major part of the molecules will be treated at the mesoscopic level. Important biological processes occur on one dimensional polymers and two dimensional membranes embedded in three dimensions. The multiscale method will be adapted to simulation of processes in cells with such internal manifolds. Space and time is discretized and for numerical accuracy the space and time steps cannot be too large but it is also known that the steps cannot be too small in the mesoscopic model and the mixed model. The restrictions on the steps in both ends will be studied.