Abordagem para guiar a reprodução de experimentos computacionais: aplicações em biologia computacional

Abstract

Systems Biology is one of the most powerful emerging areas in the third millennium that combines, in an interdisciplinary way, knowledge and tools of Biology, Computer Science, Medicine, Chemistry and Engineering. However, the continued development of computational experiments is accompanied by problems such as manual integration of tools to simulation and analysis; loss of models due software obsolescence; and the difficulty to reproduce the experiments due to lack of details of the execution environment used. Most quantitative models published in Biology are lost because they, or are no longer available or are insufficiently characterized for reproduction. This work proposes an approach to guide the registration of in silico experiments focused on its reproduction. The approach involvesthecreationofaseriesofannotationsduringcomputationalmodeling,supported by a software environment where the researcher conducts the tool integration steps, process description and execution of experiments. The goal is to noninvasively capture the modelingprocesstoincreasetheexchangeofknowledge,allowrepetitionandvalidationof the results and reduce rework in interdisciplinary research groups. A prototype was built and two different workflows have been integrated as case studies. The first uses models and tools of cardiac electrophysiology to build new applications on the environment. The secondpresentsanewuseforthesystemdynamicsmetamodelingtosimulatetheresponse of the innate immune system in a planar section of tissue. The complete capture of workflow, consisting of simulation and processing of output data, in two control experiments, was observed. The environment allowed the reproduction and adaptation of experiments at three different levels: the creation of new experiments using the same structure as the original one; the definition of new applications that use variations of the structure of the original experiment; the reuse of workflow to change models and original condition.