DISEASES

DISEASES aggregates and scores evidence for disease–gene associations by combining text-mined literature with manually curated data, cancer mutation datasets, and genome-wide association studies (GWAS).

Key Features:

• Dictionary-based named-entity recognition (NER): Uses a dictionary-based tagger to identify human genes and diseases in biomedical literature.
• Co-occurrence scoring: Applies a scoring scheme that evaluates co-occurrences within and between sentences to quantify associations.
• Evidence integration: Integrates text-mined associations with manually curated literature data, cancer mutation datasets, and GWAS results.
• Confidence scoring: Assigns confidence scores to each disease–gene association to enable comparison across evidence types.
• Performance metrics: Reported to recover approximately 50% of manually curated disease–gene associations with a false positive rate of 0.16%.

Scientific Applications:

• Hypothesis generation: Supports generation of hypotheses about genetic contributors to diseases by highlighting scored associations.
• Experimental validation: Aids validation and interpretation of experimental findings by providing integrated evidence for candidate genes.
• Target prioritization: Facilitates identification and prioritization of potential therapeutic targets based on combined evidence strength.
• Complex disease studies: Supports studies of complex diseases by aggregating diverse genetic evidence types.

Methodology:

Dictionary-based NER for human genes and diseases; co-occurrence scoring within and between sentences; integration with manually curated literature, cancer mutation datasets, and GWAS; assignment of confidence scores to associations.