DESCRIPTION:

1
ABSTRACT One of the goals of the structural
bioinformatics field is to discover how a
protein's sequence determines its structure and
how these protein structures interact with each
other. A plethora of tools that tackle specific
areas of this problem have been developed and
focus primarily on protein docking, homology
modeling and threading, structure prediction, and
similarity comparison. Unfortunately, the use of
several of these non-intuitive and user
un-friendly tools are required for the simplest
research project. The learning curve of these
tools and repetitious tasks required to operate
them consume massive quantities of researcher
time. AutomN seeks to eliminate this by
automatically wrapping many of these tools in an
easy to use GUI and automating the entire process.
Advisors Dr. Brian W. Beck1, Dr. Sergiu
Dascalu2, and Brian Westphal2 Department of
Biochemistry1, CSE Department2, UNR
Figure 2
Figure 3
Figure 1
Figure 1 AutomNs Graphical User
Interface Figure 2 Data Flow through the
AutomN Pipeline Figure 3 AutomN generated
docked structure of PDB 1A15 and 1CRN Note
AutomN is a play on the words autumn and
automatic.
DESCRIPTION AutomN is concerned with automating
the tedious task of protein interaction pathway
discovery using only protein sequences as input.
AutomN wraps many bioinformatics tools into one
easy to use, streamlined package. AutomN is
entirely GUI-based, making it so that the user
can immediately point and click to select the
options they want instead of having to consult
long manuals in order to find a specific command
line switch they desire. AutomN uses a pipeline
architecture, meaning that bioinformatics
analysis tools are called one after another on a
set of data, continually transforming it until
the desired result is achieved. AutomN takes two
protein sequences as input and attempts to locate
a structure for each sequence using BLAST. If no
structure is found, the structure is predicted
using Rosetta. Once the structure is defined for
both input proteins, they are docked together
using ZDOCK. The optimal predictions from the
ZDOCK run are then displayed to the user in a
graphical format so he or she can decide which
interactions are worth further study.
FUTURE WORK For subsequent versions, our focus
will be directed toward allowing AutomN to
utilize any bioinformatics algorithm in our
problem domain. This adaptable plug-in
architecture ensures the user is not tied to a
particular analysis module. Also, AutomN should
be able to detect if a module has failed and be
able to recover from the error gracefully.
Lastly, advanced analysis routines which help the
researcher determine the validity of the docked
structures will be added.
CONCLUSION AutomN is a new bioinformatics tool
that uses a pipeline architecture to automate the
process of converting two protein sequences into
an ensemble of probable docked protein complexes.
By understanding which proteins interact with
others, it becomes possible to modify these
interactions with newly created drugs and to
better understand the inner workings of all
organisms. The AutomN team hopes to increase
throughput of any research-oriented organization
and broaden the user base of sophisticated
research tools. In addition, AutomN is so easy
to use that almost any student can utilize it,
making it a candidate as a learning tool. The
main goals of this tool are to save time by
automating repetitive research tasks and to allow
users with limited knowledge of modeling
techniques to accomplish quite complex tasks.
For our first version, AutomN does not implement
any new bioinformatics algorithms, but focuses on
making existing algorithms easier to use. AutomN
has great potential in the research and learning
community. Hopefully with a few years of
continued work, AutomN will become commonplace in
the arsenal of widely available bioinformatics
tools.
This project was developed in Spring 2005 as part
of the course CS 426 Senior Projects