LAILAPS Search Engine (transPLANT instance)

LAILAPS Search Engine (transPLANT instance) indexes and retrieves plant genomic information to support forward genetic research by integrating millions of documents and evidence-ranked gene annotations linked to plant genome loci.

Key Features:

• Integrated Information Retrieval: Extracts knowledge from complex, heterogeneous, and distributed life science databases to provide a unified view of plant genomics.
• Multi-database Indexing: Integrates millions of documents from over 13 major life science databases, including SWISSPROT, and links evidence-ranked annotations to genome loci.
• Fuzzy Querying: Supports fuzzy queries to identify candidate genes associated with phenotypic traits across loosely integrated genome databases.
• Evidence-Based Annotation System: Associates search results with evidence-ranked gene annotations to enhance relevance and interpretability.
• Relevance Ranking via Artificial Neural Networks: Sorts search results using artificial neural networks trained on user feedback and behavior tracking to prioritize pertinent information.
• Feature Model for Relevance Discrimination: Implements a model of nine relevance-discriminating features for quantitative estimation and screening of database entries.
• User Interaction Tracking: Incorporates user behavior tracking data to refine ranking and relevance over time.
• Flexible Query Expansion: Applies a multi-stage query suggestion workflow with enhanced tokenization, word breaking, spelling correction, and query refinement to expand queries with synonyms and related terms.

Scientific Applications:

• Forward Genetic Research: Facilitates identification and retrieval of candidate genes and related evidence for mapping genotype-to-phenotype relationships.
• Gene–Trait Association Discovery and Validation: Supports retrieval and ranking of evidence-ranked gene annotations to assist discovery and validation of gene–trait associations.
• Species Case Studies (maize and barley): Demonstrated use in maize and barley for integrating and interrogating genomic information relevant to trait analysis.

Methodology:

Indexing of millions of documents from >13 life science databases including SWISSPROT; advanced information retrieval methodologies; fuzzy querying; evidence-based annotation linking to genome loci; relevance ranking via artificial neural networks trained with user feedback and behavior tracking; a nine-feature relevance-discrimination model; and a multi-stage query suggestion workflow (tokenization, word breaking, spelling correction, query refinement).