| Apoptosis is essential for normal physiology, such as development, maintenance and aging among metazoans. Drosophila melanogaster is an excellent model organism for the study of developmental apoptosis because it is genetically tractable, and the essential apoptosis mechanisms are conserved in this organism.;The validation of laboratory experiments is often done capturing images of cells and cell activity in vitro, or directly in vivo when possible. The apoptotic removal of the wing epithelium in Drosophila is a favored example because it can be easily imaged. The underlying mechanisms that govern this developmental stage are still unclear. However, the proper quantification of the apoptotic removal, especially from a spatiotemporal perspective, can aid the study of apoptosis in a population of cells, at the tissue or organ level.;Despite the striking similarity in the apoptosis pathways of fruit flies and mammals, differences have emerged. Because of these differences in the apoptosis pathway of the two species, we compiled a new model of the Drosophila apoptosis pathway. We discovered a previously unmentioned positive feedback loop between initiator and effector caspases. Through bifurcation analysis we confirmed its relationship with the characteristic all-or-none switching activity, often described with reference to apoptosis.;One of the more important issues, when modeling a population of cells, is the cell-to-cell response variability that is noticed in the population. We investigated the intrinsic stochasticity of the genetic models represented by activating and inhibitory Hill functions, using five different methods. We applied the stochastic simulation algorithm and the linear noise approximation methods to the full-scale elementary-reaction models that are expanded from the respective Hill functions. For comparison, we used the heuristic reduction methods that apply the quasi-steady-state approximation to simplify the full-network models. A more innovative method, the slow-scale linear noise approximation method was also applied. All three linear noise approximation methods show much faster computational speed when compared to the stochastic simulation algorithm counterparts. The full-scale and the slow-scale linear noise approximation compute the correct levels of intrinsic noise. The slow-scale linear noise approximation, however, achieves the same level of accuracy with considerably higher computational efficiency. |
