Data storage and AmiGO2 Working Group

Goals and Objectives:

Aim-3: Develop an online informatics portal and data warehouse for ontology-based, annotated plant genome data and plant genomes.

• Deliverables: A centralized portal for common reference ontologies for plants and the associated data sets. Novel data store and web user interface.

3.1 Planteome Web Portal Development

• Drupal portal will host the AmiGO browser, the ontology database (similar to the one developed by the PO and the GO), and a BioMart
• Transition to AmiGO 2.0 with new features

3.2 Planteome Data Warehouse Development

• Novel data warehouse for storing both the ontologies and annotation data based on NoSQL (e.g. MongoDB, http://www.mongodb.org, and Apache™ Hadoop®, http://hadoop.apache.org)
• Integrate the MapReduce algorithm to increase scalability and performance
• Investigate using HDF (Hierarchical Data Format), as a storage format for any numerical or sequence-based data.
• Create an efficient way to add annotations incrementally to the database, (not possible in the current AmiGO database)
• Implementation of OLAP (Online Analytical Processing) data cubes (http://en.wikipedia.org/wiki/OLAP_cube)

3.3 Integration with the iPlant infrastructure

• Initial design and testing will happen locally at the Center for Genome Research and Biocomputing at Oregon State University
• Use of virtual machine (VM) images in the iPlant cloud computing environments
• Utilization of high performance computing resources, such as:
  • The supercomputer 'Stampede' at Texas Advanced Computing Center (TACC)
  • Use of iRODS at iPlant for data file storage and retrieval
  • Image hosting via Bisque hosted on the iPlant infrastructure (See 3.4, below)
• Interaction with resources such as CoGE, Bisque, and the Integrated Breeding Platform (IBP)

3.4 Library of Publicly-Accessible, Annotated Digital Images

• Design a relational data schema to support the large-scale storage of annotated images (and their associated metadata)
• Image library main goal: A training set for a new auto-segmentation and annotation active-learning algorithm
• Support other visual analysis tools and the integration of image data with ontology data
• Will also function as a home for community-contributed image data

3.5 Application Programming Interface (APIs)

• Develop of publicly available APIs for both internal and external data access to ontology terms and annotations
• Extend the existing lightweight web services providing Plant Ontology terms, synonyms, and definitions to the Planteoem APIs, including direct web service access to annotated data
• Potential Users:
  • Gramene project- information about annotations and ontologies
  • DOE KBase project (http://kbase.science.energy.gov/)
  • iPlant tools and services.

• Integrate our data with other external APIs, For example:
  • EBI (the Gene Expression Atlas, Ensembl Plants, IntAct),
  • ERA-CAPS (genotype-to-phenotype data)
  • DOE KBase
  • GCP Integrated Breeding Platform
  • Agave on iPlant which provides web-focused developer access to the iPlant data store and other integration services, providing a direct link to high-performance computing systems such as the TACC.

Participants

• Jaiswal Lab (OSU, BPP): Justin Elser
• Mungall Group (Lawrence Berkeley National Laboratory): Chris Mungall (Co-PI), Seth Carbon
• Zhang Lab (OSU, EECS): Eugene Zhang (Co-PI), Botong Qu (CS Ph.D. student)