WebApollo: A Web-Based Sequence Annotation Editor for Community

Annotation

Ed Lee, Gregg Helt, Nomi Harris, Mitch Skinner, Christopher Childers, Justin

Reese, Jay Sandaram, Christine Elsik, Ian Holmes, Suzanna Lewis

Bioinformatics Open Source Conference (BOSC 2011)

July 15, 2011

“Old” ApolloPopular open source genome annotation editing

toolLets biologists view computational analyses and

experimental evidence for genomic features and build a manually-curated consensus

Standalone Java application

In The Old Days

Users had to download and install Apollo

Annotations saved locally in flat files

Sharing done by emailing files

Only one person could edit a region at a time

Apollo in the Old Days

(Flat file)

Starting To Get Better

Annotations saved directly to a centralized databaseEdits made by other users not visible until you

actively reloadedPotential issues with stale annotation data

Apollo software downloaded more transparently by Java Web Start

Users still needed to install Java

Starting To Get Better

WebApollo:Collaborative Annotation

Web-based (runs in browser)

No software download required

Annotations saved to centralized databaseEdit server mediates annotation changes made by

multiple users

Real-time annotation updating

WebApollo Framework

Web-based Client

Uses JBrowse Javascript annotation browserFastHighly interactive

WebApollo extensions to JBrowseGestures for editing annotationsCommunication with the annotation editing engine

and data providing serviceHTML5 Canvas quantitative data rendering

Multiple Client Synchronization

Comet modelServer pushes annotation updates to all clients

in real time

Annotation Editing Engine

Java

Handles all the logic for editing

Edits stored persistently in the serverBerkeleyDB JE for fast accessAble to restore data if either client or server

crashes

Per-sequence (contig, chromosome, etc.) user permissions (none/read/write)

Enabling WebApollo to access as many types of genomic data as

possible

Efficient access to public data from UCSC, Ensembl, and GMOD Chado databases

Unified strategy (like DAS) is preferred

Solution: Trellis, a DAS server framework that: Supports modular back-end plugins, e.g.,

Poka Trellis plugin for UCSC database access

DAS plugin for Ensembl servers

Supports modular front-end content formats, e.g., JBrowse JSON plugin

Trellis Framework Servlet

Poka Plugin

Trellis

DAS Data Model

DAS Model to UCSC MySQL

MySQL response to DAS Model

UCSC MySQL

Genome database

WebApollo Client

JBrowse JSON Plugin

Model to JBrowse JSON

DAS HTTP Request with format modifier

WebApollo DemoTry it yourself:

http://icebox.lbl.gov:8080/ApolloWebDemo

Annotations you make on demo server are:Not kept privateNot guaranteed to persist

Source Code(BSD License)

Client source codehttps://github.com/berkeleybop/jbrowse

Annotation editing enginehttp://code.google.com/p/apollo-web

Data model and I/O layerhttp://code.google.com/p/gbol

Trellis server codehttp://code.google.com/p/genomancer

Acknowledgements

LBNL Ed Lee Gregg Helt Nomi Harris Suzanna Lewis

UC Berkeley Mitch Skinner Ian Holmes

Georgetown University Christopher Childers Justin Reese Jay Sundaram Christine Elsik

Demo: http://icebox.lbl.gov:8080/ApolloWebDemo