Differential Transcript Usage (DTU)

DTU detects proportional shifts in transcript isoform expression from the same gene locus. While differential gene expression (DGE) might show no net change, DTU reveals that a gene produces a different dominant isoform—potentially encoding a protein with altered function. This captures alternative splicing, alternative promoter usage, and alternative polyadenylation events that DGE completely misses.

DTU analysis typically employs generalized linear models with a Dirichlet-multinomial or beta-binomial distribution to account for transcript-level overdispersion. Tools like DEXSeq, DRIMSeq, and SUPPA2 model transcript proportions rather than raw counts, testing for significant interaction effects between condition and isoform usage while controlling for gene-level expression changes.

DTU reveals critical regulatory events including:

• Exon skipping: Cassette exons included or excluded
• Intron retention: Transcripts retaining intronic sequence
• Alternative 5'/3' splice sites: Shifted exon boundaries
• Alternative first/last exons: Promoter or polyadenylation site switching These events frequently drive tissue-specific expression, developmental transitions, and disease pathogenesis.

Aberrant isoform switching is a hallmark of many cancers and neurological disorders. DTU analysis identifies disease-specific transcript isoforms that may serve as highly specific biomarkers or therapeutic targets. For example, the CD44 gene switches from standard to variant isoforms during epithelial-mesenchymal transition, a process DTU directly quantifies while gene-level analysis shows minimal change.

A critical analytical nuance: DTU tests whether transcript usage fractions change between conditions, independent of overall gene expression. This requires distinct statistical machinery—DGE tools like DESeq2 test absolute abundance changes, while DTU tools test compositional shifts within a gene's transcript pool. Confounding these two analyses leads to false positives where apparent DTU is merely a reflection of gene-level differential expression.

DTU results are commonly visualized using proportion plots showing transcript usage fractions per condition, sashimi plots from tools like IGV or ggsashimi displaying junction read coverage, and heatmaps of isoform-level expression. Interpretation requires integrating transcript annotations from Ensembl or GENCODE to assess whether isoform switches alter coding sequences, untranslated regions, or regulatory elements.

DTU identifies oncogenic isoform switches that promote tumorigenesis. For example, the PKM2 isoform of pyruvate kinase is preferentially expressed over PKM1 in many cancers, shifting metabolism toward aerobic glycolysis (the Warburg effect). Similarly, alternative splicing of BCL-X produces the anti-apoptotic BCL-XL isoform, which confers chemoresistance. DTU analysis of tumor-vs-normal RNA-seq data systematically detects these therapeutically targetable switches without requiring prior knowledge of the splicing event.

DTU captures the precise isoform repertoires that define tissue identity. The troponin T (TNNT2) gene expresses mutually exclusive exons that fine-tune calcium sensitivity in cardiac versus skeletal muscle. Fibronectin (FN1) includes the EDA and EDB exons in embryonic tissues but excludes them in adult liver. Comparing DTU across tissue panels reveals these developmental and cell-type-specific splicing programs, providing a molecular fingerprint of differentiation state.

The brain exhibits the highest rates of alternative splicing, and DTU is essential for mapping isoform dynamics during neural development. The Dscam gene in Drosophila can generate over 38,000 isoforms through alternative splicing, encoding cell-surface receptors critical for neuronal self-avoidance. In mammals, activity-dependent splicing of neurexin (NRXN) genes alters synaptic adhesion properties, directly modulating neurotransmitter release probability. DTU analysis links these isoform transitions to specific developmental stages and plasticity events.

DTU reveals how alternative splicing dynamically reshapes the proteome during immune responses. Upon T-cell activation, CD45 switches from the RABC isoform to the shorter RO isoform, altering phosphatase activity and lowering the signaling threshold. The CD44 gene shifts from the standard CD44s isoform to variant isoforms (CD44v) containing additional extracellular exons, which modulate lymphocyte homing and are implicated in autoimmune disease. DTU captures these rapid, condition-dependent isoform transitions.

DTU identifies alternative splicing events that mediate drug resistance, a major challenge in targeted therapy. Resistance to vemurafenib in BRAF-mutant melanoma involves a splicing switch in BRAF itself, producing a truncated isoform that dimerizes in a drug-insensitive manner. In prostate cancer, alternative splicing of the androgen receptor (AR) generates constitutively active AR-V7 variants that drive resistance to enzalutamide. DTU analysis of pre- and post-treatment biopsies systematically detects these escape mechanisms.

DTU tracks programmed isoform switches that drive developmental progression. The fetal-to-adult hemoglobin switch involves silencing of HBG1/HBG2 (γ-globin) and activation of HBB (β-globin), a transition targeted therapeutically for sickle cell disease. In myogenesis, MEF2C transcription factor isoforms shift from a repressive to an activating form, controlling muscle differentiation. DTU analysis of time-course RNA-seq data precisely maps the kinetics of these essential developmental isoform transitions.

The biological mechanism that generates multiple mature mRNA isoforms from a single gene. Key modes include:

• Exon skipping: The most common form in mammals, where an exon is excluded
• Intron retention: An intron remains in the mature transcript
• Alternative 5'/3' splice sites: Shifting the boundary of an exon
• Mutually exclusive exons: Only one of two exons is included DTU analysis quantifies the functional consequences of these splicing decisions.

A specific DTU pattern where the dominant transcript isoform for a gene changes between conditions, even if overall gene expression remains constant. This often indicates a functional shift—for example, switching from a membrane-bound receptor isoform to a soluble decoy receptor. Tools like IsoformSwitchAnalyzeR integrate functional annotation to predict the biological consequences of such switches.

The computational process of reconstructing full-length mRNA isoforms from short-read RNA-seq data. Tools like StringTie and Salmon perform reference-guided assembly, while Trinity handles de novo assembly. Accurate transcript assembly is a critical prerequisite for DTU analysis, as isoform-level quantification depends on resolving the complete set of transcripts expressed in a sample.

A cellular surveillance pathway that degrades mRNA transcripts containing premature termination codons. Many alternative splicing events introduce PTCs, targeting the isoform for NMD. DTU analysis can reveal regulated unproductive splicing, where a gene's expression is tuned by producing NMD-targeted isoforms that are degraded rather than translated into protein.