WS-PGRADE/gUSE-Based Science Gateways in Teaching - Science Gateways for Distributed Computing Infrastructures - page 220

scripting languages to access high-performance computing resources might not be
adequate, projects of this scale targeted for this audience are only possible to realize
using science gateways. Additionally, the heterogeneity of used formats and
docking protocols of the different tools, namely Autodock Vina, FlexX, and
CADDSuite (Trott 2010; Rarey 1996; Kohlbacher 2012) make the comparison a
tedious endeavor. Since the variety of different docking protocols can be repre-
sented ef ciently by workflows, they can be utilized to guarantee uniform data

handling and reasonable comparability of simulation results. In this way the stu-
dents are able to focus on the scienti

c questions without getting lost in technical
details. The requirements to successfully carry out such projects are provided by the
MoSGrid portal, a WS-PGRADE/gUSE-based science gateway for molecular
simulations (see Chap. 11 ) (Gesing 2012; Kacsuk 2011a), MoSGrid exposes a
powerful distributed computing infrastructure and presents resources to the end user
in an easy-to-operate graphical user interface. Additionally, the science gateway
relieves the students from the nontrivial task of compiling and installing the indi-
vidual docking tools and other technical dif culties, in order to allow the student to

mainly focus on the scienti c challenges. The computing power is kindly provided

ü ü

by the Zentrum f Tübingen.

In order to create a demanding task, each group was required to perform the data
analysis completely on their own, using various scripting languages such as R or
Python. The goal was to create meaningful visualizations of the results, e.g., charts
and plots illustrating the quality of docking results. This included scatter plots
(experimental vs. simulation), evaluation of binding poses (root mean square
deviation of the center of mass of ligands), enrichment factors (actives vs. inac-
tives), and receiver

für Datenverarbeitung (ZDV) of the University of T

operator curves (true positives rate vs. false positives rate). The

-

cate enabling access
to the computing resources connected to MoSGrid. Although the provision of
certi

first assignment for the students was to obtain a personal certi

cates is well organized and documented, not all students succeeded in
obtaining one. Despite the fact that there were no technical or bureaucratic diffi- -

culties in obtaining a personal certi cate, we have noticed through several years of

teaching work that some students might struggle in meeting deadlines. Considering
that obtaining a personal certi

cate takes up to a week, students were encouraged to
apply as early as possible. The application for a personal certi cate was part of the

first weekly assignment together with a small introductory task facilitating the
capabilities of MoSGrid to familiarize the students with science gateways and to
prevent failure of students at later stages.
The next steps included the preparation of the input

files and the docking of the

prepared libraries, which was facilitated by the filtering, selection and docking

capabilities of the docking portlet (Fig. 16.3 ). Nevertheless, some basic under-
standing of the protein docking target is required to be able to make appropriate
selections.
Due to technical dif

finish their
projects. In general, a portal should be able to report which of its components is not
working properly. For common problems, gUSE-based portals and MoSGrid in
particular, comprehensive error capturing is in place. But due to the complex

culties, some student groups could only partly