The study team applied PacBio HiFi long-read sequencing to six human left ventricular samples: two non-failing, two with dilated cardiomyopathy (DCM), and two with ischemic heart disease (IHD). The investigators curated a catalogue of 48,232 isoforms across 11,760 genes. Among these, 15,535 were annotated as novel NIC (novel in catalog) and NNC (novel not in catalog) isoforms. The authors described this as an isoform-resolved reference of the human heart in health and disease. The manuscript is a preprint and therefore has not completed peer review.
Short-read re-quantification and additional cohort analysis were part of the downstream strategy described in the abstract; the source did not report detailed sample-level demographics, sequencing depth per sample, or additional wet-lab parameters in the abstract text.
Data indicated independent evidence for most novel structures. For isoforms that contained novel splice junctions, 81.7% had short-read support for all novel junctions. The authors also cross-referenced 5′ ends with a cardiac CAGE atlas and reported support for 93% of NIC and 86% of NNC isoforms. These layers of orthogonal evidence—long-read structure, short-read junction confirmation, and CAGE-defined transcription start sites—were used to increase confidence in the annotated novel transcripts.
The study team explicitly reported those validation percentages in the abstract; the source did not provide, within the abstract, the detailed computational thresholds used to call a junction supported or the precise CAGE datasets and matching criteria.
Authors observed that novel isoform diversification was concentrated in sarcomere genes. This enrichment remained after the investigators adjusted for potential confounders that could produce apparent diversity—specifically gene complexity, expression level, sequencing depth, and transcript length. The abstract states that sarcomere-related genes showed the strongest novel isoform signal, implying that functionally important cardiac structural genes harbor previously unannotated transcript diversity.
The source abstract did not list individual sarcomere genes or provide gene-level examples; these details may appear in the full manuscript or supplementary material but are not present in the abstract text.
To assess disease-associated transcript changes, the investigators augmented GENCODE v43 with 18,637 non-FSM long-read isoforms. They then re-quantified two independent short-read cohorts—MAGNet and GSE116250—using the expanded reference. The authors emphasized that many transcript-level changes in DCM exceeded what was visible at the gene level. Specifically, for most genes carrying a transcript that changed more than twofold in DCM, the corresponding gene-level change was less than twofold. This indicates that gene-centric expression analysis can mask substantial isoform shifts.
Across cohorts, 626 transcripts were differentially expressed in both datasets; 98% of those transcripts changed in the same direction in each cohort. The abstract notes that all 100 long-read isoforms that were among those differentially expressed showed concordant directional change between cohorts. The source did not specify in the abstract the statistical thresholds, correction methods, or sample counts used for differential expression testing beyond the cohort names and fold-change criteria reported.
The authors made the isoform resource available through an interactive portal called CardioIsoMap (link provided in the preprint). The abstract identifies CardioIsoMap as an isoform-centric portal intended to allow users to explore the cardiac isoform atlas. The resource is described as a comprehensive, isoform-resolved reference of the human heart in both health and disease based on the long-read dataset and augmented annotations.
The abstract did not include usage details for the portal, programmatic access methods, or file-level download instructions; interested users should consult the preprint and supplementary materials for those specifics.
The document is a preprint and has not been certified by peer review. The abstract does not report certain methodological specifics: it omits sample demographic details, per-sample sequencing coverage metrics, precise computational pipelines and thresholds, and lists of individual novel isoforms or gene-level examples. Those items appear necessary to evaluate reproducibility and will require consultation of the full manuscript and supplementary files. The authors did not report regulatory milestones or peer-review status beyond the preprint posting.
Open questions that remain—based on what the abstract reports versus what it omits—include the extent to which the novel isoforms are translated, whether protein-level validation exists for sarcomere isoforms, and how representative six samples are of broader DCM and IHD populations. The source did not provide answers to these points in the abstract; readers should refer to the full preprint and CardioIsoMap for more complete methods and data.