Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Forecasting promising biological simulations at PhysiBoSS

Anesti Effrosyni

Full record

---

Summary

Since the existing biological multicellular systems are characterized by high complexity and heterogeneity, coupled with the fact that there has been a remarkable upsurge in computer science, in-silico methods based on mathematical models are in a great use. Specifically, they are particularly helpful when we must deal with diseases that have abnormal and unpredicted behavior, such as cancer or auto-immune ones. The need for understanding and curing such kind of diseases, led us to the integration of different modelling frameworks that take into account the intra- and the extra-cellular environment, as well as, the interplay between cells. Such an example is PhysiBoSS, that combines two other well-established frameworks to support its whole functionality and provide us a cell-fate decision model with an accurate representation of cells’ population variance through the time under a specific treatment and conditions.Considering the fact that not all PhysiBoSS simulations are hopeful, to facilitate the procedure of results’ collection and examination, the bad simulations must be excluded. This thesis’ goal is to design a distributed and parallel system that implements a forecasting algorithm on a great amount of real-time running simulations and decides about the sustainability or not of a simulation’s execution and finally maintains only the top k hopeful ones out of all the initial simulations. The algorithm’s performance was evaluated both locally and remotely/distributively, giving us very positive results.

Available Files

Services