IsoTag: A Universal Assembly-Free System for Deterministic Identification of Transcript Isoforms

The advent of long-read RNA sequencing has revealed vast numbers of previously unannotated transcript isoforms. However, the field lacks a standardized system to identify and track these unannotated isoforms across studies, assemblies, and databases. Here we present IsoTag, a universal and assembly-free identification framework that deterministically encodes transcript structures and sequence variants. IsoTag extends the GA4GH Variation Representation Specification (VRS) to transcripts, employing SHA-512-based cryptographic hashing to generate reproducible, globally unique identifiers. Applied to large-scale long-read RNA-seq data, IsoTag achieved memory-efficient processing, cross-assembly consistency, and seamless integration into existing BAM/SAM workflows via standardized that (i) preserves the isoform Structure (XI) and sequence Variant (XV) identifiers, and (ii) introduces reversible tags Splice tag (XS) and Boundary tag (XB) that encode exon junctions and transcript ends as compact hexadecimal coordinates decodable without a genome FASTA. A Transcript group (XT) identifier supports sample-independent biological clustering with fuzzy 5′/3′ boundaries. By providing the first universal transcript identifier system, IsoTag addresses a foundational barrier to reproducible transcriptomics, enabling scalable integration across single-cell studies, population datasets, and reference transcript annotations. open-source (MIT license): https://github.com/LSBDT/isotag