Computer science solutions for molecular biology problems are often presented in the form of workflows. There is a set of activities performed by different processing entities through managed tasks. Knowledge about the data trajectory throughout a given workflow enables reproducibility by data provenance. In order to reproduce an in silico bioinformatics experiment one must consider other aspects besides those steps followed by a workflow. Indeed, the computational settings in which the involved programs run is a requirement for reproducibility. Cloud computing technology may hide the technical details and make it easier for the user to set up such an on-demand environment. NoSQL database systems have also gained popularity, particularly in the cloud. Considering this particular scenario, we have planned and executed a research study about a bioinformatics workflow running in an IaaS cloud computing environment. We have persisted provenance data according to the PROV-DM model, using different types of NoSQL database systems. We present in this paper some preliminary results from our research work, where we have explored the characteristics of several NoSQL database systems to persist provenance data.
Many scientific experiments in Bioinformatics are executed as computational workflows. Frequently, it is necessary to re-run an experiment under the original circumstances in which it was run to recognize and validate it. Data provenance concerns the origin of data. Knowing the data source facilitates the understanding and analysis of the results, by detailing and documenting the history and the paths of the input data, from the beginning to the end of an experiment. Therefore, in this context, data provenance can be applied when experimenting traceability. This document presents AProvBio, an architecture that can perform the data provenance of scientific experiments in bioinformatics automatically, using the provenance data model PROV-DM and in a graph database. The architecture can perform the automatic provenance type prospectively, retrospectively and with user-defined data. Thus, the architecture stores and captures information obtained during the execution of the data generation processes with user-defined data information, such as features and versions of the programs used. A graph model, based on the PROV-DM model, was proposed for storing the data provenance. The PROV-DM can be represented by a graph, it allows for a more natural modelling, as well as expressing queries at a more natural level, and the implementation of efficient algorithms to perform specific operations.