RatsPub

RatsPub mines PubMed sentences for co-occurrence of user-specified gene symbols and addiction-related keywords to identify gene–addiction relationships.

Key Features:

• Sentence-level literature mining: Identifies sentences that contain both user-specified gene symbols and over 300 addiction-related keywords organized into six categories.
• Keyword ontology: Uses a set of >300 addiction-related keywords organized into six distinct categories for targeted searches.
• PubMed access and retrieval: Queries the NIH PubMed server via a programming interface and retrieves relevant abstracts from a local copy of the PubMed archive.
• Convolutional neural network classification: Employs a convolutional neural network to distinguish sentences describing systemic stress from those describing cellular stress.
• GWAS integration: Integrates the NHGRI-EBI GWAS catalog to link literature findings with human GWAS results.
• GeneCup ontology integration: Integrates with the GeneCup framework for searches across PubMed abstracts using custom keywords organized into a customized ontology.

Scientific Applications:

• Gene–addiction relationship discovery: Identifies and aggregates literature evidence for associations between specific genes and addiction-related concepts.
• GWAS and omics synthesis: Facilitates integration of NHGRI-EBI GWAS catalog results and other omics findings with the biomedical literature.
• Stress-context disambiguation: Refines interpretation of stress-related literature by separating systemic stress mentions from cellular stress mentions.

Methodology:

Queries the NIH PubMed server using a programming interface, retrieves abstracts from a local copy of the PubMed archive, mines sentences for co-occurrence of user-specified gene symbols and >300 addiction-related keywords organized into six categories, employs a convolutional neural network to classify systemic versus cellular stress sentences, and integrates NHGRI-EBI GWAS catalog entries and GeneCup keyword ontology.