Will Computers Crash Genomics
SCIENCE VOL 331 11 FEBRUARY 2011
R01945014 黃博強 R01945037 林彥伯 R01945039 蘇醒宇 R01945043 吳卓翰 R01945046 蘇煒迪 R01945017 陳維
Introduction
Old Genome Informatics
The Evolution of DNA Sequencing
New Genome Informatics
Dizzy with data
Dizzy with data
• Human Genome Project
– Planned for 15 years
• Celera Genomics
– Shotgun Sequencing Method
Shotgun Sequencing Method
Assemble fragments
Assemble fragments
Dizzy with data
• After 2005
– Sequence generation
– Ability to handle the data
• “Next-generation” machines – Cheaply
– Faster
• Computer – Memory
– Processing
Dizzy with data
• Genome Project – More
• Third generation machines
– Smaller
Storage Issues
Cost v.s.Data
3.2 billion base pairs X 1,000 X 10,000 = USD$ 32,000,000
USD$
3,200
Problems facing Bioinformatic
Data storage Data transfer
Data Storage
• Bioinformatics field tend to archive all raw sequence data.
More than 90 GB
Data Transfer
• Want to analyze a genome?
More than 594 GB
Solving the problem (storage)
• Discard the original image files , an d only keep the sequence data.
• If necessary, just re-sequence the sa
mple.
Solving the problem (storage)
• Putting the data in an off-site facil ity.
$0.095 per GB-month of data stored (Singapore)
$0.100 per GB-month of data stored (Tokyo)
$0.500 - $1.000 per GB of data stored
Solving the problem (transfer)
• Put one copy of the data in the common c loud which everyone uses.
• Encouraged by the genomics community
– NCBI
• has put a copy of the data from the pilot project of the 1000 Genomes effort into off-site storage.
– Ensemble, the EBI sequence database
• are automatically funneled into a cloud environme nt as part of a test of the strategy.
Worries about security
• Data involving the health of human subjects , which is being linked more and more to ge nome information
• The Health Information Protection Regulatio ns came into force on July 22, 2005.
– The Health Information Protection Act is designe d to improve the privacy of people’s health inf ormation while ensuring adequate sharing of info rmation is possible to provide health services.
Going To the Cloud
• National Human Genome Research Institute(NHGR I) hosted several meetings on cloud computing and on informatics and analysis in 2010.
• “One thing that is clear is that as computat ion becomes more and more necessary through- out biomedical research, the way these [infra structure] resources are funded will have to change to be more efficient,” says James Ta ylor, a bioinformaticist at Emory University
Growing Exponentially of Da
ta
• The primary goal of bioinformatics is to increase the understanding of biol ogical processes
• But “We live in the post-genomic era , when DNA sequence data is growing e xponentially“
Miami University (Ohio) computational biologaist Iddo Friedberg
NCBI Data Growth
EMBL Data Growth
grand area of research
• Sequence analysis
• Genome annotation
• Analysis of gene expression
• Analysis of protein expression
• Analysis of mutations in cancer
• Protein structure prediction
• Comparative genomics
• Modeling biological systems
• High-throughput image analysis
• Protein-protein docking
• Sequence analysis
– most primitive operation in computationa l biology
• Genome annotation
– the process of marking the genes and oth er biological features in a DNA sequence
• Analysis of gene expression
– The expression of many genes can be dete rmined by measuring mRNA levels
• Analysis of protein expression
– Gene expression is measured in many ways including mRNA and protein expression
• Analysis of mutations in cancer
– to identify previously unknown point mut ations in a variety of genes in cancer
• Protein structure prediction
– important for drug design and the design of novel enzymes
• Comparative genomics
– the study of the relationship of genome structure and function across different biological species
• Modeling biological systems
– a significant task of systems biology an d mathematical biology
• High-throughput image analysis
– Computational technologies are used to accelerat e or fully automate the processing, quantification and analysis of large amounts
• Protein-protein docking
– predict possible protein-protein interactions based on 3D shapes
Obstacles in Computing Tech
nology
Two Ways to Approach higher Computin g Ability
• One Computer Computing Ability
• Cloud Computing
One Computer Computing Ability
• TSMC 20nm manufacture procedure
• No direct co-relation of bus observed data with th e internal CPU activity
• Multi-core processor : record and replay (R&R) sys tem
Intel Corporation:
Virtues and Obstacles of Hardware-assisted Multi-processor Execution Replay (2010)
Cloud Computing
• Availability of a Service
• Data Lock-in
• Data Confidentiality and Auditability
• Data Transfer Bottlenecks
• Performance Unpredictability
• Scaling Quickly
“10 Obstacles To Cloud Computing” By UC Berkeley & How GoG rid Hurdles Them