Transcriptome profiling provides the ultimate molecular phenotype, revealing signatures of genetic activity that are characteristic to specific tissues, cells and disease subtypes. Advances in RNA sequencing technologies have drastically enhanced biomedical research, facilitating the discovery of new cell types, tissue organization and the identification of novel diagnostic and prognostic biomarkers. However, RNA-seq is fraught with artifacts–-both technical and analytical–-substantiating the need for synthetic spike-in controls, rigorous benchmarking and biological replicates. Indeed, most RNA-seq experiments (including single-cell and spatial applications) are restricted to gene-level quantifications that rely on (often incomplete) reference transcriptome annotations, ignoring alternative splicing isoforms and cell-type specific long non-coding RNAs. Third-generation sequencing technologies herald a new era for molecular medicine, where long reads from native molecules are generated in real-time. We assessed the clinical utility of RNA-seq with nanopore for the diagnosis of paediatric Acute Lymphoblastic Leukemias (ALL) from bone marrow aspirates using a gene-level neural network classifier trained on heterogeneous short-read sequencing data. In addition to successfully classifying 11/12 representative ALL samples, this strategy was able to accurately classify disease subtypes in as little as 3 minutes of sequencing, enabling a same-day molecular diagnosis of ALL for less than CA$200. We hypothesize that isoform-level transcriptomes will greatly increase the transcriptomic resolution of complex diseases and facilitate the identification of new biomarkers. Consequently, we have undertaken rigorous bioinformatics benchmarking of long-read de novo transcriptome assembly pipelines using Sequins synthetic RNAs to establish a gold standard modus operandi for third-gen transcriptomics. To demonstrate the power of this approach, we performed de novo transcriptome assembly from a leukemia sample using long reads produced from the popular 10X Chromium single-cell platform to identify thousands of new transcript isoforms, which increased single-cell analysis granularity when compared to Illumina sequencing data alone.