Alternative splicing (AS) is a key gene regulatory process that enables individual genes to produce multiple transcript isoforms, substantially increasing transcriptomic and proteomic diversity. Aberrant AS has been linked to many diseases, including cancer; however, transcriptomic analyses commonly prioritise gene-level expression and therefore fail to capture the regulatory complexity introduced by isoform variation.
Recent developments in single-cell long-read RNA sequencing (SCLRS) make it possible to directly profile full-length transcript isoforms in individual cells, opening new avenues for investigating transcript diversity across cell populations, tissue contexts, and disease settings. Despite this progress, computational approaches designed specifically to analyse and interpret isoform-level complexity at single-cell resolution remain underdeveloped.
Our research program responds to this gap by developing computational and machine learning methods for isoform-resolved transcriptomic analysis. These include tools for alternative splicing quantification (SpliceWiz), single-cell isoform complexity analysis (ScIsoX), and machine learning approaches for identifying splicing regulators and classifying cell types from isoform expression profiles. ScIsoX, in particular, offers a systematic framework for measuring transcriptomic complexity through new isoform-level metrics and visualisation strategies.
Here, we introduce ScIsoX, a computational framework for systematically characterising isoform-level transcriptomic complexity in single cells. ScIsoX combines a hierarchical data structure, a set of complexity metrics, and purpose-built visualisation tools to quantify transcript diversity driven by alternative splicing. This framework supports the investigation of global and cell-type-specific complexity patterns, uncovering multidimensional transcriptomic signatures that conventional gene-centred analyses may overlook.