Dataflow processing in a global networked context: a solution for the computational methods pool management problem

摘要

While the number of computational methods available to approach problems in computational chemistry is steadily increasing, synergical effects from the combination of a multitude of complementary methods fail to materialize in the expected extent. We introduce a system which provides an extensible networked environment for the computation, analysis and visualization of chemical data of arbitrary definition. The core is a flexible data manager connected to a network of algorithmic module databases. Upon accepting a request for some piece of chemical information, the system looks up methods to derive this information and autonomously invokes the necessary procedures to obtain the requested property data. Other types of modules available in a similar fashion include data analysis procedures or tools and visualization components. Complex sequences of computations and data analysis are specified in a visual programming environment, using building blocks obtained from sites all over the world. An increase in new algorithm implementation efficiency and notable synergical effects are expected.