galaxy: integrative, reproducible analysis of genomics data
DESCRIPTION
Galaxy: Integrative, Reproducible Analysis of Genomics Data. Genomic and Proteomic Approaches to Heart, Lung, Blood and Sleep Disorders Jackson Laboratories Ross Hardison September 10, 2008 Galaxy is developed and maintained by Anton Nekrutenko (PSU) and James Taylor (Emory U). - PowerPoint PPT PresentationTRANSCRIPT
Galaxy: Integrative, Reproducible Analysis of Genomics Data
Genomic and Proteomic Approaches to Heart, Lung, Blood and Sleep Disorders
Jackson LaboratoriesRoss Hardison
September 10, 2008
Galaxy is developed and maintained by Anton Nekrutenko (PSU) and James Taylor (Emory U)
Types of data in genomics
• Sequences • Comparisons of DNA and protein sequences• Expression data• Chromosomes and chromatin data• Experimental manipulation• Variation and phenotypes• Protein structure and function• Stored in databases and browsers (e.g. UCSC
Genome Browser)• Many analysis tools (Galaxy)
Some major web resources in genomics
• UCSC Genome Browser and Table Browser– http://genome.ucsc.edu/
• Ensembl and EnsMart/BioMart– http://www.ensembl.org/
• TIGR Comprehensive Microbial Resource – http://cmr.tigr.org/
• NCBI for Blast server, PubMed, Gene Expression Omnibus, dbSNP, etc. – http://www.ncbi.nlm.nih.gov/
• dCode for alignments and other – http://dcode.org
• HapMap for haplotype and variation– http://hapmap.org
• Galaxy for data retrieval and analysis– http://galaxy.psu.edu
Sequences
• DNA sequences – Whole genomes and chromosomes– Genes
• Transcripts– Protein-coding and noncoding transcripts– Full-length or partial (expressed sequence tags or
ESTs)• Protein sequences
– Known– Predicted
• Repeats• Variants
Sequences from CFTR: Browser view
Regulation-related features around T2D risk variants
Reg Pot
Browsers vs Data Retrieval
• Browsers are designed to show selected information on one locus or region at a time.– UCSC Genome Browser– Ensembl
• Run on top of databases that record vast amounts of information.
• Sometimes need to retrieve one type of information for many genomics intervals or genome-wide.
• Access this by querying on the tables in the databases or “data marts”– UCSC Table Browser– EnsMart or BioMart
Retrieve all the protein-coding exons in humans
Challenges in genomic data analysis
• We have great browsers and data warehouses– But most lack facilities for performing
sophisticated analysis
• Many useful computational tools have been developed in bioinformatics– But they are not well integrated, they have
different user interfaces, different data formats, etc.
Some common solutions
• Glue it all together with Excel– Until you realize Excel cannot handle that
much data and the match isn’t coming out right anyway…
• Glue it all together with Perl– But that leads to duplication of effort,
duplication of bugs, ….
A better solution
• Build a framework that:– Defines a common format for describing the
interfaces of different computational tools and databases
– Provides the infrastructure to adapt those interfaces into standard form
– Defines common data types and standards for integrating the results
Two faces of Galaxy
• A web site where you can easily perform complex analysis integrating various data sources and computational tools
• A framework to easily build similar sites that integrate your choice of tools and data sources
Galaxy: Data retrieval and analysis
• Flexible data retrieval– From multiple external sources– Upload from user’s computer– Upload as URL from any site
• Hundreds of computational tools– Data editing, filter, sort– File format conversion– Extract sequences and alignments– Operations: merge, intersection, complement, cluster …– Get conservation and other scores for intervals– Statistics– Graphs and displays– EMBOSS tools for sequence analysis– HyPhy tools for molecular evolutionary analysis
• Workflows: run multiple steps reproducibly
Welcome to Galaxy
Welcome screen, changes periodically
News
When tools are invoked, displays information on the tool and allows user to chose parameters
Tool choice
Titles are toggles; more options are displayed when you click on them
History
Titles are toggles; more information is displayed when you click on them
Click on the “eye” to see all the data on another page
Click on the “pencil” to edit the attributes
Click on the “x” to delete
“Refresh” to get results if they have not appeared or to get status of query
Use “options” next to “History” to save, rename, move to or share histories. Must be logged in to do this.
Proxy based tools (e.g. UCSC Table Browser)
User makes request to Galaxy
Galaxy delegates request to external site
Proxy based tools
External site generates response - If data, Galaxy determines data type, processes it and adds it to the history - Otherwise, response is returned to user
Command line tools
Pick one of the programs from the left “Tools” bar
User chooses parameters for tool
Command is run
Background jobs in Galaxy
Web page with datasets on transcriptional regulation
Data uploads to Galaxy: use the URL
How many DHS overlap with high RP intervals?
Overlaps of DHS with high RP
segments (25%) and
highly constrained segments
(43%)
24,330/95,709 = 0.25441,000/95,709 = 0.428
Get constraint scores for intervals
Histogram of phastCons scores
Mean vs Maximum phastCons
Distribution of phastCons scores in DHS that are also occupied by CTCF
mean max
n=7000
Many thanks …
James Taylor, Anton Nekrutenko,
Funding from NIDDK, NHGRI, Huck Institutes of Life Sciences at PSU
Yong Cheng, Demesew Abebe, Christine Dorman, …, Ying Zhang, David King, Swathi Ashok Kumar