RNA sequencing workflows increasingly rely on transcriptome-based alignment as it offers superior isoform-level quantification accuracy and computational efficiency compared to genome-based approaches. However, aligners that map reads to a transcriptome reference produce BAM files with transcript-relative coordinates, which cannot be directly used for genome browser visualisation, variant calling, RNA editing detection, or integration with genome-based analyses. Converting these alignments back to genomic coordinates typically requires re-alignment with splice-aware aligners, which is computationally expensive and may yield inconsistent alignment decisions. The growing adoption of direct RNA sequencing (DRS), whose long reads span multiple exons and carry high error rates, further compounds this challenge. A fast, accurate, BAM-level converter supporting both short- and long-read data has remained unavailable.
We present T2G, a tool that converts transcriptome-aligned BAM files to genomic coordinates. T2G uniquely handles both short, ultra-short footprinting, and long-read RNA-seq datasets, addressing the distinct challenges of each technology. The tool builds comprehensive transcript models from GTF/GFF annotations, assigns read positions and reconstructs genomic CIGAR strings while preserving tag information. T2G manages strand reversal, sequence reverse-complementation, quality reversal for negative-strand genes, and enriches output files with extensive metadata including gene and transcript identifiers, genomic span metrics, intron-spanning indicators, gene-relative positions, and preserved modified-base and polyA tail tags.
On an ONT DRS dataset, T2G retained 100% of the 5,725,956 transcriptome-mapped reads after liftover to genomic coordinates. The resulting splice junctions were 99.9% canonical with expected acceptor and donor sequence pairs, compared with only 72% canonical for reads aligned directly to the genome. Validation across diverse transcriptomes, including human, rat, and yeast datasets demonstrated accurate coordinate conversion that scales across genome complexity.
The effective BAM-level conversion of T2G with preservation of read tags bridges transcriptome-based quantification and genome-based analysis to enable comprehensive genome-wide RNA-seq workflows.