PubGene

PubGene constructs an automated gene-to-gene co-citation network from over 10 million MEDLINE records to reveal relationships among 13,712 named human genes.

Key Features:

• Automated Knowledge Extraction: Employs advanced algorithms to extract explicit and implicit biomedical knowledge from MEDLINE titles and abstracts across more than 10 million records.
• Gene-to-Gene Co-Citation Network: Generates a co-citation network that maps relationships between genes based on literature co-occurrence, encompassing 13,712 named human genes.
• Annotation with MeSH and GO Terms: Links each gene to relevant Medical Subject Headings (MeSH) and Gene Ontology (GO) terms to provide functional and contextual annotations.
• Validation through Large-Scale Experiments: Validated in three large-scale experiments demonstrating that gene co-occurrence in the literature reflects biologically meaningful relationships.
• Gene Expression Analysis of Microarray Datasets: Applied to analyze two publicly available microarray datasets to demonstrate applicability in gene expression investigations.

Scientific Applications:

• Hypothesis Generation: Facilitates generation of hypotheses by identifying literature-supported gene associations.
• Literature-Based Discovery: Enables discovery of novel relationships through analysis of co-citation patterns in the biomedical literature.
• Exploration of Therapeutic Targets: Supports exploration of potential therapeutic targets by highlighting annotated gene associations relevant to diseases and processes.
• Contextualization of Gene Function: Provides contextualization of biological processes, molecular functions, and cellular components via MeSH and GO annotations.
• Interpretation of Gene Expression Data: Assists interpretation of microarray gene expression datasets by mapping expression changes onto literature-derived gene networks.

Methodology:

Applies advanced algorithms to extract information from MEDLINE titles and abstracts, constructs a co-citation network from gene co-occurrence data, links genes to MeSH and GO terms, and validates results using three large-scale experiments and analysis of two public microarray datasets.