IEEE Women in Engineering Magazine - June 2017 - 15

types of data "clean up" algorithms
based on how the data was used in a
workflow that reduced the data footprint and investigated optimal ways to
configure cloud resources to maximize
overall performance.

Impacting Science
Deelman and her team translated
their research into a software product
that is being used by many science
applications. The Pegasus Project
that she designed at USC in collaboration with the HTCondor team at
the University of Wisconsin-Madison,
researches how to conduct automated
processes in distributed systems. It
looks at data management, task and
job scheduling, issues of reproducibility, and how to use cloud computing, clusters, and HTCondor pools.
The software tools make CS research In 2016, Deelman received the USC/ISI
Achievement Award for "technical contributions
principles available to scientists.
and leadership in the field of scientific workflow
"The Pegasus Workflow Manage- systems for high-performance computing."
ment tool allows scientists to describe
computations in an abstract way,"
amounts of data to confirm the detecDeelman explains. "For example, if you
tion and wanted to use other resources.
run an application on this data, the
Pegasus allowed them to perform their
result will be another data set that can
computations and seamlessly use other
be processed by another app. Providresources such as the Open Science
ing Pegasus with a graph or a recipe
Grid and XSEDE. It offered a
of what to do, it looks at the
huge benefit. They didn't
computational resources
have to rework their workand data available, maps
There is an
flows or figure out how
the computations to
additional need
to use these different
the resources, and
resources; Pegasus did
plans the movement
for people with
of the data across difinterdisciplinary skills, it for them."
"It managed the
ferent resources."
combining CS and
movement of data
Pegasus has helped
domain science.
computation, monimanage large-scale
tored how computations
computations and data
and data stages progress,
involved in the Laser Interand allowed them to recover
ferometer Gravitational-Wave
from certain failures, Deelman continObservatory (LIGO) Gravitational Wave
ues. "LIGO workflows may have milphysics project since 2001. Collaborating
lions of tasks. LIGO scientists don't
with computer scientists, astronomers,
want to babysit every single failure. Pegand gravitational physicists, she helped
asus allows them to recover without losimprove their computer systems and data
ing data or computations. If a task fails,
processing in the computational environit resends it to another resource to try
ment that led to the first detection of a
another system to do the computation.
gravitational wave.
It automates the tedious tasks."
"LIGO had its own infrastructure
Pegasus is helping bioinformafor computing and storing data," she
ticians unlock the secrets of cancer
says. "But they needed to process large

June 2017

cells. It was used with the Montage
astronomy software to verify the
presence of a new structure called
a bar in the spiral galaxy M31. And
it helped generate the world's first
physics-based probabilistic seismic hazard map for the southern
California region by efficiently
managing workflow tasks in the
CyberShake project.
Deelman also applied automation
technologies such as Pegasus to perform quality control on data used in
a study linking adipose and insulin
biology to body fat distribution. In
addition, she participates in large
cyberinfrastructure projects such as
the Open Science Grid and XSEDE
and was part of the first National
Science Foundation cloud project
FutureGrid. "Pegasus is a terrific
example of a project that looks at
how people can work together to
build systems that enable science to
be done at large scale," she says.

Helping People
There's an idea that computer scientists
sit behind a computer all day, but Deelman gets to work with people, joining
in on large-scale projects, collaborating, and sharing ideas. In addition to
working with other computer scientists,
she is in contact with people in various science communities-astronomy,
bioinformatics, climatology, earthquake
science, gravitational-wave physics,
material science, and others.
"Working in STEM fields is rewarding; you get challenging problems and
the opportunity to collaborate with people," Deelman says. "That's something
that doesn't come through about CS
often. Computer science is everywhere.
There are great chances to make a real
impact anywhere."
Deelman adds that there is an additional need for people with interdisciplinary skills, combining CS and domain
science. She advises anyone interested in
joining the CS field to take on broad collaborations between different domains or
specialize in one domain and go deeper.
-Debbie Sniderman

IEEE womEn In EngInEErIng magazInE

Table of Contents for the Digital Edition of IEEE Women in Engineering Magazine - June 2017