Page 65 - profile2012.indd

Research Faculty 研究人員

研究員 研究員
宋定懿 Ting-Yi Sung 李丕榮 PeiZong Lee

Research Fellow Research Fellow
Ph.D., Operations Research, New York University Ph.D., Computer Science, New York University

Tel: +886-2-2788-3799 ext. 1711 Fax: +886-2-2782-4814 Tel: +886-2-2788-3799 ext. 1812 Fax: +886-2-2782-4814
Email: tsung@iis.sinica.edu.tw Email: leepe@iis.sinica.edu.tw
http://www.iis.sinica.edu.tw/pp./tsung/eindex.html http://www.iis.sinica.edu.tw/pages/leepe

● Research Fellow, IIS, Academia Sinica (2000~) ● MBA, State University of New York at Buffalo (1983) ● Research Fellow, Academia Sinica, Taiwan (1998/1–present)
● Associate Research Fellow, IIS, Academia Sinica (1989~2000) ● BS, Management Science, National Chiao Tung University (1980) ● Associate Research Fellow, Academia Sinica, Taiwan (1989/10–1998/1)
● Ten Outstanding Young Women Award (1998)
● Ph.D., Operations Research, New York University (1989)

Research Description Publications Research Description Publications

My research interest is bioinformatics in proteomics, and 1. Hsin-Nan Lin, Cédric Notredame, Jia-Ming Chang, Ting-Yi My research interests are in compilers for scienti c appli- 1. PeiZong Lee, Chih-Hsueh Yang, and Jeng-Renn Yang, “Fast

speci cally, protein structure prediction and mass spec- Sung, Wen-Lian Hsu, Improving the alignment quality of cations, parallel algorithm design, computer architectures, Algorithms for Computing Self-Avoiding Walks and Mesh

trometry (MS)-based proteomics analysis. consistency based aligners with an evaluation function using and the interplay among architectures, algorithms, and Intersections over Unstructured Meshes,” Advances in Engi-
synonymous protein words, PLoS ONE, volume 6(12), pages neering Software, volume 35, number 2, pages 61--73, Febru-

In protein structure prediction area, we have developed e27872, December 2011. compilers. The challenge of implementing large scienti c ary 2004, (Elsevier Science B. V.)
methods for secondary structure prediction with high 2. Allan Lo, Cheng-Wei Cheng, Yi-Yuan Chiu, Ting-Yi Sung, applications for current parallel computers is to handle 2. PeiZong Lee and Wen-Yao Chen, “Generating Communica-
performance. We further focused on prediction of mem- Wen-Lian Hsu, TMPad: an integrated structural database for data locality among memory hierarchies, so that to avoid tion Sets of Array Assignment Statements for Block-cyclic

brane protein structures since membrane proteins are helix-packing folds in transmembrane proteins, Nucleic Acids memory con ict on shared memory parallel computers Distribution on Distributed Memory Parallel Computers,”
Parallel Computing, volume 28, number 9, pages 1329--1368,
prominent drug targets and few of their structures have Research, vol. 39, Suppl. 1 (Database issue), pp. D347-355, and to avoid communication overhead on distributed September 2002, (Elsevier Science B. V.)
been experimentally determined due to experiment dif- 2011. memory parallel computers. Compilers act as bridges con- 3. PeiZong Lee and Zvi M. Kedem, “Automatic Data and Com-
culties. We have developed e cient methods for pre- 3. Chih-Chiang Tsou, Chia-Feng Tasi, Ying-Hao Tsui, Putty- necting algorithms and architectures. I am interested in putation Decomposition on Distributed Memory Parallel

dicting transmembrane helices and topology, helix-helix Reddy Sudhir, Yi-Ting Wang, Yu-Ju Chen, Jeou-Yuan Chen, studying this interdependence. Computers,” ACM Transactions on Programming Languages
interaction and contact, lipid exposure, rotational angles, Ting-Yi Sung, Wen-Lian Hsu, IDEAL-Q: An automated tool For running on distributed memory parallel computers, if and Systems, volume 24, number 1, pages 1--50, January 2002.

classi cation of transmembrane helices. We have also con- for label-free quantitation analysis using an efficient peptide a program does not use indirect memory accesses, such 4. Chih-Hsueh Yang, PeiZong Lee, Yeh-Ching Chung, “Improv-

structed a knowledge base for all known helix-helix inter- alignment approach and spectral data validation, Molecular as subscript array of arrays or pointers, compilers can nd ing Static Task Scheduling in Heterogeneous and Homogene-

ous Computing Systems,” Proceeding International Confer-
and Cellular Proteomics, vol. 9, no. 1, pages 131-144, 2010.
actions in currently available structures. We will integrate data dependence relations among statements and data ence on Parallel Processing, Xi-An, China, September 2007.
all of our prediction methods to further pursue tertiary 4. Hsin-Nan Lin , Ching-Tai Chen , Ting-Yi Sung , Shinn-Ying alignment relations among data arrays. Therefore, compil- 5. Chih-Hung Hsu and PeiZong Lee, “Study of Flow Calcula-
structure prediction of membrane proteins. Ho and Wen-Lian Hsu, Protein subcellular localization predic- ers have enough information to determine data distribu- tions at All Speeds on Unstructured Grids,” Proceedings 14
th
tion of eukaryotes using a knowledge-based approach, BMC National Computational Fluid Dynamics Conference, Nantou,
Liquid chromatography combined with mass spectrome- Bioinformatics, volume 10, Suppl. S15, pages S8, December tion, execution scheduling, and the generation of commu- Taiwan, August 2007.

try has become a predominant technology for large-scale 2009. nication code using e ective communication primitives. 6. PeiZong Lee, Chih-Hsueh Yang, and Jeng-Renn Yang,
proteomics research. The goal of MS-based proteomics is 5. Allan Lo, Yi-Yuan Chiu, Einar Andreas Rødland, Ping-Chiang However, if a program does use subscript array of arrays “Repartitioning Unstructured Meshes for the Parallel Solution

to identify di erentially expressed proteins in di erent cell Lyu, Ting-Yi Sung, and Wen-Lian Hsu, Predicting helix-helix or pointers to implement indirect memory accesses for of Engine Combustion,” Supercomputing’2003, Phoenix, AZ,
states, e.g., tumor vs normal cells. As we have developed a interactions from residue contacts in membrane proteins, Bio- irregular computation, compilers at the compiling time November 2003.
suite of automated tools for protein quantitation analysis, informatics, volume 25(8), pages 996-1003, February 2009. cannot determine neither data dependence relations nor 7. PeiZong Lee, Chih-Hsueh Yang, and Jeng-Renn Yang, “Fast
we have shifted our research focus to protein identi ca- 6. Chih-Chiang Tsou, Yin-Hao Tsui, Yi-Hwa Yian, Yi-Ju Chen, data alignment relations; if compilers only provide naïve Algorithms for Computing Self-Avoiding Walks and Mesh

th
Intersections over Unstructured Meshes,” 16 AIAA Compu-
tion. As reported in the literature, only a small portion, typ- Han-Yin Yang, Chuan-Yih Yu, Ke-Shiuan Lynn, Yu-Ju Chen, data distributions, generated code cannot avoid adopt- tational Fluid Dynamics Conference, Orlando, FL, June 2003,
ically about 30%, of all acquired spectra in an experiment Ting-Yi Sung, and Wen-Lian Hsu, MaXIC-Q Web: a fully ing expensive communication primitives which de nitely AIAA paper 2003-4125

automated web service using statistical and computational

were identi ed by existing identi cation tools. Further- methods for protein quantitation based on stable isotope labe- lengthen the execution time and thus degrade the perfor- 8. PeiZong Lee, Jan-Jan Wu, and Chih-Hao Chang, “Partitioning

more, since current tools can identify even less modi ed ling and LC-MS, Nucleic Acids Research, volume 37, Suppl 2 mance of parallel processing. Unstructured Meshes for Homogeneous and Heterogeneous
proteins, we have worked on improving identi cation of (Web Server issue), pages W661-W669, 2009. To understand the techniques for compiling irregular Parallel Computing Environments,” Proc. International Con-

proteins with some types of modi cations, e.g., phospho- 7. Jia-Ming Chang, Emily Chia-Yu Su, Allan Lo, Hua-Sheng computation, we analyze real code for scienti c computa- ference on Parallel Processing, pages 315--322, Vancouver,

British Columbia, August 2002.

rylation, nitrosylation and biotinylation, and will tackle Chiu, Ting-Yi Sung, and Wen-Lian Hsu, PSLDoc: protein tion, in which we have to conduct research into unstruc-

more types of modi ed proteins. In addition to MS-based subcellular localization prediction based on gapped-dipep- tured mesh generation; unstructured mesh partition; 9. PeiZong Lee, Chih-Hao Chang, and Jan-Jan Wu, “Parallel
Implicit Euler Solver on Homogeneous and Heterogeneous
proteomics, we are currently working on MS-based me- tides and probabilistic latent semantic analysis, PROTEINS: Computing Environments,” 15 AIAA Computational Fluid
th

tabolomics to develop tools for quantitation and identi - Structure, Function, and Bioinformatics, volume 72, number Euler equation and Navier- Stokes equation solvers for a Dynamics Conference, Anaheim, CA., June 2001, AIAA pa-
cation of metabolites. 2, pages 693-710, February 2008. numerical wind tunnel platform, an engine combustion per 2001-2588
platform for computing reactive ows; visualization; and 10. PeiZong Lee, Chih-Hao Chang, and Maw-Jyi Chao, “A Paral-

the challenge of using MPI on workstation/PC clusters to lel Euler Solver on Unstructured Mesh,” Proc. ISCA 13 Inter-
th
accelerate irregular computation. national Conference on Parallel and Distributed Computing
Systems (PDCS-2000), pages 171--177, Las Vegas, Nevada,
August 2000.
研究人員
64 Research Faculty
65