PolyA_DB

PolyA_DB catalogs polyadenylation sites (PAS) in mammalian genomes using transcriptomic data to enable identification, annotation, and comparative analysis of alternative polyadenylation isoforms.

Key Features:

• Comprehensive PAS identification: The database contains approximately 1.4 million PAS in both human and mouse genomes, representing 4.9- and 3.5-fold increases in coverage over the previous version for each species respectively.
• 3'READS+ data integration: PAS are identified through exhaustive mining of approximately 2.3 billion PAS-supporting reads per species generated by the 3' region extraction and deep sequencing plus (3'READS+) method.
• Long-read RNA sequencing validation: 20% of the full PAS set matches transcript end sites (TES) from public long-read RNA sequencing (LR-RNA-seq) data, which substantially complements RefSeq-based gene assignment and enables APA event subtyping by splicing configuration.
• PAS conservation and strength metrics: The database provides quantitative information on PAS conservation across species and relative PAS strength.
• UCSC Genome Browser integration: Database contents are linked to the UCSC Genome Browser for genomic visualization of PAS locations and features.

Scientific Applications:

• Alternative polyadenylation isoform annotation: Researchers can identify and characterize APA isoforms resulting from multiple PAS usage in protein-coding and long non-coding RNAs.
• 3' end artifact detection: Comparison between annotated PAS and LR-RNA-seq TES enables identification of potential 3' end artifacts derived from internal A-rich regions of RNA.
• APA-splicing relationship analysis: LR-RNA-seq data integration supports subtyping of APA events in the context of splicing configuration to understand coupled RNA processing mechanisms.
• Comparative genomics: PAS conservation information enables cross-species comparison of polyadenylation regulatory landscapes in mammalian systems.

Methodology:

PAS identification involves exhaustive mining of human and mouse transcriptomic data sets generated by 3'READS+, with validation against public long-read RNA sequencing data and RefSeq annotation for gene assignment.