SparkINFERNO

SparkINFERNO infers molecular mechanisms of non-coding genetic variants by integrating GWAS summary statistics with functional genomics datasets to prioritize causal variants, regulatory elements, tissue contexts across more than 400 tissues and cell types, and potential target genes.

Key Features:

• Integration of functional genomics data: Integrates GWAS summary statistics with enhancer activity, transcription factor binding, expression quantitative trait loci (eQTLs) and other functional genomics datasets across more than 400 tissues and cell types.
• Causal variant prioritization: Identifies and prioritizes causal variants underlying GWAS association signals and annotates implicated regulatory elements and putative target genes.
• Scalability and computational implementation: Implements an API using Apache Spark for parallel processing and scalable handling of large genomic datasets.
• Genomic indexing and performance: Employs Giggle-based genomic indexing and has been reported to achieve over 60-fold efficiency improvements compared to traditional methods on large datasets.
• Evaluation on large GWAS studies: Has been evaluated on large-scale GWAS studies for efficient processing and high-accuracy identification of causal variants and their molecular mechanisms.

Scientific Applications:

• Non-coding GWAS interpretation: Dissects molecular mechanisms of non-coding GWAS variants by linking variants to regulatory elements and tissue-specific contexts.
• Target gene nomination: Prioritizes candidate target genes affected by non-coding variants to inform downstream functional studies of complex traits and diseases.
• Therapeutic hypothesis generation: Supports generation of mechanistic hypotheses and candidate therapeutic targets relevant to precision medicine and disease biology.

Methodology:

Integrates GWAS summary statistics with functional genomics datasets (enhancer activity, transcription factor binding, eQTLs) using an Apache Spark–based parallel architecture with an API and Giggle-based genomic indexing.