exscitech: expanding volunteer computing to explore science, technology, and health

Post on 25-Dec-2014

531 Views

Category:

Education

1 Downloads

Preview:

Click to see full reader

DESCRIPTION

M. Matheny, S. Schlachter, L. Crouse, E. Kimmel, T. Estrada, M. Schumann, R. Armen, G. Zoppetti, and M. Taufer: ExSciTecH: Expanding Volunteer Computing to Explore Science, Technology, and Health. In Proceedings of the 2nd workshop on Analyzing and Improving Collaborative eScience with Social Networks (eSoN 12), October 2012, Chicago, Illinois, USA.

TRANSCRIPT

ExSciTecH: Expanding Volunteer Computing to Explore Science, Technology, and Health

M.  Matheny1,  S.  Schlachter1,  L.M.  Crouse2,  E.T.  Kimmel2,    T.  Estrada1,  M.  Schumann3,  R.  Armen3,  G.  ZoppeB2,  and    

M.  Taufer1  

1University  of  Delaware  2University  of  Millersville  

3Thomas  Jefferson  University  

Volunteer Computing •  Volunteer Computing (VC) is a

form of distributed computing in which volunteers donate processing and storage resources to computing projects.

•  Every job represents a portion of a larger problem whose computation is divided into smaller chunks and addressed in parallel.

•  Applications suited for VC include searches in very large spaces, parameter tuning, and data analysis.

1

Volunteers •  Are not representative of the general population

•  White males with a background in computers •  Paradigm causes them to be passive •  Lose interest in the project and uninstall the VC software within a few

months of participation

2

Main Contribution •  Problem: Volunteer Computing appeals to a very narrow demographic

–  We want to utilize intuitive technologies and user interfaces to appeal to historic minorities in Science, Technology, Engineering and Mathematics (STEM)

•  Problem: In VC participants are generally passive and not involved in the research process –  We want to get volunteers to help solve important research problems

and make scientific discovery

3

Outline •  Background •  Motivation and Plan •  Implementation •  Game Testing •  Conclusion and Future Work

4

Outline •  Background •  Motivation and Plan •  Implementation •  Testing •  Conclusion and Future Work

5

Docking@Home

6

•  Docking@Home is powered by Berkley Infrastructure for Network Computing (BOINC)

D@H Scientific Goals •  Self-Docking

–  Searching for protein inhibitors –  Targeted diseases

•  HIV •  Breast cancer (trypsin)

•  Cross-Docking –  Identifying drug side effects –  Look at proteins similar to the disease protein –  Naïve approach: all to all

•  All proteins •  All ligands

–  This is where we want to utilize volunteers to reduce search space

7

Related Work •  Fold.it

–  Developed by the same team as Rosetta@Home –  Aims to use volunteers to fold proteins through puzzles –  Complements VC system, does not integrate with VC system

•  Bossa –  Developed by the BOINC team –  Volunteers use cognition, knowledge, and intelligence to solve

problems •  Luis von Ahn’s work

–  Phetch – improving web accessibility –  Games with a purpose (GWAP) –  reCAPTCHA

8

Outline •  Background •  Motivation and Plan •  Implementation •  Testing •  Conclusion and Future Work

9

Docking@Home: As it stands

10

ExSciTecH: Learning Stage

11

ExSciTecH: Engaging Stage

12

ExSciTecH: Engaging Stage

13

ExSciTecH: the full circle

14

Outline •  Background •  Motivation and Plan •  Implementation •  Testing •  Conclusion and Future Work

15

BOINC Infrastructure

16

BOINC"

Local File System"

Web Browser"

Upload"

Download"

BOINC DB"

Daemons"CGI

BOINC"BOINC"Client"

Web Interface!

Front-end!Client!

BOINC!

C L I E N T" S E R V E R"

Back-end!

BOINC + ExSciTecH

17

Player" Teacher" BOINC"

Game DB"

Local File System"

Web Browser"

Upload"

Download"

BOINC DB"

Daemons"CGI

BOINC"

CGI Learn"

CGI Engage"

BOINC"Client"

Learn"Game"

Engage Game"

Web Interface!

Front-end!Client!

Player!Teacher!BOINC!

C L I E N T" S E R V E R"

ExSciTecH"Daemons"

Back-end!

VMD"

ExSciTecH Client

•  Modularly designed – each game is a separate program and the client acts as a manager

18

Learning Games •  Teach volunteers about science without intimidating them •  Several levels available:

–  High school student –  Under graduate student –  Graduate student –  Pharmaceutical student –  Professional chemist

•  As players progress through the levels the game becomes more challenging

•  Familiarize volunteers with the science •  Give D@H more exposure

–  Students in classrooms

19

Molecule Flashcards

•  Players must identify or categorize a 3D molecule as it flies towards them •  If the player incorrectly identifies the molecule they’re given access to

additional information about it

20

Engaging Games •  Volunteers have been trained by the learning stage •  Building a job

–  Short game –  Game objects correlate to job input parameters

•  Protein (disease) •  Ligand (drug) •  Ligand Confirmation •  Ligand Rotation

•  Submitting a job –  Game submits these parameters to the server –  Server builds a job based on these parameters

•  Getting results –  Good results More games! –  Volunteering CPU time More games!

21

Drag’n’Dock

22

•  Volunteers rotate a protein and select a ligand

•  Then they fly a spaceship over the protein with the ligand in tow

•  They attempt to dock the ligand in the protein with the space ship

•  The game submits data to the server to build a D@H work unit

Drag’n’Dock •  Volunteers rotate a

protein and select a ligand

•  Then they fly a spaceship over the protein with the ligand in tow

•  They attempt to dock the ligand in the protein with the space ship

•  The game submits data to the server to build a D@H work unit

23

Drag’n’Dock •  Volunteers rotate a

protein and select a ligand

•  Then they fly a spaceship over the protein with the ligand in tow

•  They attempt to dock the ligand in the protein with the space ship

•  The game submits data to the server to build a D@H work unit

24

Outline •  Background •  Motivation and Plan •  Implementation •  Testing •  Conclusion and Future Work

25

Testing the Molecule Flashcard Game •  What we want to learn:

–  Do people learn more with the game than they do with a paper test? –  Do they enjoy our game more than a paper test? –  What improvements can we make to the game?

•  How we tested: –  We took a group of students and had half of them attempt to identify

molecules with our flashcard game and half of them attempt to identify molecules on a paper test

•  What we measured: –  Time to complete the game/test –  Number of molecules correctly identified –  Survey with level of enjoyment and comments

26

Testing Setup •  24 computer science students

–  14 undergraduate students –  10 graduate students

27

Vs.

•  10-minute introduction •  Students split into two groups

–  Paper test –  Molecule flashcard game

Results and Discussion

28

•  Make more errors, but enjoy the game more •  Tend to be faster – less reflection

Results and Discussion

29

•  Learning Curve •  Confusing 3D representations

Outline •  Background •  Motivation and Plan •  Implementation •  Testing •  Conclusion and Future Work

30

Conclusions and Future Work •  We can transform the way volunteers participate in VC projects

–  More accessible –  More exciting

•  We show students had a higher level of enthusiasm when using ExSciTecH rather than traditional learning tools

•  We identified improvements that could be made to the flashcard game: –  Variable speed –  Pause –  Skip and come back

•  We are moving forward with the ExSciTecH development by continuing development of engaging games

31

Acknowledgments

32

Thanks  to:  M.  Matheny  (UD)  S.  Schlachter  (UD)    L.M.  Crouse  (U.  Millersville)  E.T.  Kimmel  (U.  Millersville)  T.  Estrada  (UD)  M.  Schumann  (TJU)  R.  Armen  (TJU)  G.  ZoppeB  (U.  Millersville)  

Sponsors: Contact:  taufer@udel.edu  

GCLab  group  

IIS #0968350/#0968368

top related