Cells In Silico (CIS): A biomedical simulation framework based on Markov random field

This paper presents CIS, a biomedical simulation framework based on the markov random field (MRF). CIS is a discrete domain 2-D simulation framework emphasizing on the spatial interactions of biomedical entities. The probability model within the MRF framework facilitates the construction of more realistic models than deterministic differential equatio n approaches and cellular automata. The global phenomenon in CIS are dictated by the local conditional probabilities. In addition, multiscale MRF is potentially useful for the modelling of complex biomedical phenomenon in multiple spatial and time scales. The methodology and procedure of CIS for a biomedical simulation is presented using the scenario of tumor-induced hypoxia and angiogenesis as an example. The goal of this research is to unveil the complex appearances of biomedical phenomenon using mathematical models, thus enhancing our understanding on the secrets of life. Computational cell biology is an emerging discipline where biomedical simulations are employed for the study of cells and their microenvironments in various spatio-temporal scales. The E-cell and the Virtual Cell projects focus on the molecular and biochemical level within cells, addressing the dynamics of signal transductional, regulatory and metabolic networks. The sub-cell compartmental model are constructed and integrated gradually so as to simulate a particular facet (or pat hway) of cells. The Epitheliome project is an example of tissue-level simulation, aiming to depict the epithelial cell growth and the social behavior of cells in culture. Simulations on higher-level systems include Physiome, and the modelling of many organs such as heart. Each scale of simulation shed light on different aspects of life. Biomedical simulations have been conducted in both the continuous and discrete domains. Differential equations are the key elements of continuous domain simulation, where the concentration of particular receptors, ligands, enzymes or metabolites are modelled at various spatial and temporal scales. This approach is limited by the fact that many biomedical phenomena are too complex to be described by sets of differential equations. In addition, the deterministic differential equations are not adequate for describing many biological phenomenon with a stochastic nature. Alternatively, discrete domain simulation are processed on a spatio-temporal discrete lattice. T he combination of Pott’s model and Metropolis algorithm have been used to simulate cell sorting, morphogenesis, the behavior of malignant tumor and the Tamoxifen treatment failure of cancer.