12 Igra 4pm

Repository Development Strategy: 

Repository Development Strategy Mark Igra, Fred Hutchinson Cancer Research Center migra@fhcrc.org

Overview: 

Overview Where we are now Goals and challenges for the repository Lessons from successful projects Our development strategy

Mouse & Sample Tracking: 

Mouse & Sample Tracking

MS2 Storage & Analysis: 

MS2 Storage & Analysis

A typical MS2 workflow: 

A typical MS2 workflow Sample Digest Digested Sample Fractionate Digestion Protocol Fractionation Protocol chromatogram Fractions Online LC Mass Spec Pump pressure etc Protocol MS Output Settings Search Matches Search options Curation Results Depletion Depletion Protocol Depleted Sample Rules?? Matches MS Output

Goals & Challenges: 

Goals & Challenges Goals Build a repository that people want to use Incorporate experiments and samples from many different labs Help establish widely used standard Challenges Experimental process is a moving target Must integrate existing software We need the repository now

History of the Web: 

History of the Web Invented by Tim Berners-Lee, November, 1990 Very Simple New Technologies URLs, HTML, HTTP Reuse lots of old technology TCP/IP, MIME types Standards committees came after it was a working popular system Many focused standards handled by different bodies (IETF, W3C, ISO, ECMA)

Why the Web Succeeded: 

Why the Web Succeeded Low Barriers to entry Platform & language independent Simple, human readable file formats Free & open tools available early “Evolvable” Extensible Loosely Coupled Decentralized Partial support yields partial (but useful) results Useful to lots of people Not just programmers Provided value to both data providers and data consumers

Linux: Cathedral and Bazaar: 

Linux: Cathedral and Bazaar Essay from Eric Raymond Open development process + Open Source When it works Starting point from individual or small group Show “plausible promise” that system will be great Release early & release often Rapid public release/test/improve cycle Still have leadership that decides what goes in But they are committed to open process & inclusion

Our Strategy: 

Our Strategy Start with an extensible framework Pick reasonable starting point for annotation Provide free, open tools for annotation Refine through real-world use Make it easy for community to participate

Extensible Annotations: RDF: 

Mouse Extensible Annotations: RDF Everything has unique id and properties that describe it Web Ontology Language (OWL) for schema/ontology sample:bdi/011404-12 1/27/05 individual:kemp.fhcrc.org/mouse/AA-28 Liver 11/22/04 http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#Mouse Individual date-taken species tissue-type Birth_Date http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#Tissue http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#Date http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#Organisms http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#Birth_Date http://mged.sourceforge.net/ontologies/MGEDOntology.owl#Individual http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#Liver

Tools for Annotation: 

Tools for Annotation Sample Annotator Ship samples with electronic manifest Experiment Annotator Rapidly describe an experiment Make reuse easier than reinvention But allow for new invention Still need “push” systems customized for each lab But share components and designs

Slide13: 

Experiment Overview Proteins of Interest Samples Assays

Summary: 

Summary Concentrate on an evolutionary model Open source and open process Build tools for people What we need Data contributors Code contributors

Extra Slides: 

Extra Slides

Extensible Annotations: RDF: 

Extensible Annotations: RDF RDF (Resource Description Framework) is a general purpose information representation system Describe anything using a set of “statements” Subject: id of the thing we are talking about Predicate: property we are describing Object: value of that property Statements generally combined in XML files Web Ontology Language (OWL) describes schema/ontology Combine your own properties or reuse other people’s NCICB already publishes Ontology in OWL format

RDF Graphically: 

RDF Graphically sample:bdi/011404-12 1/27/05 individual:kemp.fhcrc.org/mouse/chodosh/AA-28 Liver 11/22/04 http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#Mouse Individual date-taken species tissue-type Birth_Date http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#tissue-type http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#Date http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#Organisms http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#Birth_Date http://mged.sourceforge.net/ontologies/MGEDOntology.owl#Individual

Why RDF: 

Why RDF Very general information representation system Not a constrained world – can describe or refer to items outside of current domain Good open libraries of tools Simple to reuse existing ontologies MGED ontology for samples

What makes a great standard?: 

What makes a great standard? What attributes generate adoption? What language do you use to describe the standard What social model do you use to “invent” What social model do you use to “standardize” A great standard is one that people want to use because it makes their life better

Folksonomy as a model for experiment annotation: 

Folksonomy as a model for experiment annotation Folksonomy – an emergent classification system based on the input of groups of users rather than on top-down structure (see taxonomy) del.icio.us bookmark organizing site is prime example Users post bookmarks with “tags” describing them View bookmarks of any user View bookmarks by term Find relationships between terms by clustering how they are used together

Social models for creating & refining standards: 

Social models for creating & refining standards Dictator Central Committee Anarchy (no standard) The fourth way – open source model Cathedral and the Bazaar Release early and release often Leadership

A typical MS1/MS2 workflow: 

A typical MS1/MS2 workflow Mass Spec MS Output Settings Feature Identification Matches Search options Alignment Results Alignment Options Feature List MS Output

A typical MS2 workflow: 

A typical MS2 workflow

Assay View: 

Assay View

Application to Experimental Annotation: 

Application to Experimental Annotation Experiments are designed independently May include novel techniques & processes However, many concepts & processes are shared Expose existing process descriptions (yours & other people’s) Make reuse easier than reinvention

Experiment Builder: 

Experiment Builder Graphical tool to build an experiment description Allows you to string together a set of actions into a process Can reuse existing processes Rapidly find previously used options Yours & other peoples