AutoDock Gateway for Molecular Docking Simulations in Cloud Systems - Cloud Computing with e-Science Applications - page 218

Molecular docking simulations are a widely utilized application area in
which parameter sweep scenarios are desired. Molecular docking simulation
packages, for example, AutoDock [ 1 ], are applied by various disciplines, such as
molecular biology, computational chemistry, or even psychology, and require
a large number of cloud resources to increase the speed of computation.
To collect the required number of cloud resources, end users like biologists
and chemists would have to learn the cloud interfaces. However, instead of
learning such information technology (IT) systems, they would rather like to
concentrate on their own scientific field and research. To hide this low-level
technology from them, high-level user interfaces like science gateways are
required. WS-PGRADE [ 2 ] was designed with this idea to provide high-level
graphical workflow abstraction for users to hide the low-level details of access-
ing the underlying cloud infrastructures. WS-PGRADE provides workflow tem-
plates that tremendously simplify the creation of parameter sweep (and other
types of workflow) applications and takes care of accessing the required type
and number of cloud resources. To achieve this, WS-PGRADE was integrated
with the CloudBroker Platform (CBP), which enables accessing heterogeneous
cloud resources, including Amazon EC2, IBM SmartCloud, Eucalyptus [ 3 ],
OpenStack [ 4 ], and OpenNebula [ 5 ] clouds. Moreover, WS-PGRADE also sup-
ports the development of intuitive and customized end-user interfaces to com-
pletely hide the underlying complexity from the scientist.
This chapter demonstrates how parameter sweep application scenarios,
such as AutoDock-based molecular docking experiments, on cloud computing
infrastructures can be efficiently supported by the WS-PGRADE framework
that is the core technology of the European Union FP7 project SCI-BUS [ 6 ],
which is aimed at developing various science gateways for a large set of dif-
ferent scientific user communities.
10.2 WS-PGRADE Workflows and Parameter Sweep Application
WS-PGRADE workflows are represented as directed acyclic graphs
(DAGs), for which nodes denote computational tasks (or some other work-
flow), and directed arcs denote data dependency between the different
nodes. Figure 10.1 shows a workflow with six nodes, where different data
dependencies are defined. When submitting this workflow, the first two
nodes ( Copy_A and Copy_B ), as they have no prerequisites, are able to run
immediately on the targeted computing infrastructure. Once Copy_A has
finished and produced the necessary input for the Invert_A node, this job
is ready for submission. Similarly, once Copy_B and Invert_A have both
finished, WS-PGRADE can start processing the Multi_B node. That is, once
all the preceding nodes of a given workflow node have finished successfully,
the given node is run by WS-PGRADE.