Beyond the gene: contextualizing novel isoforms through pathway analysis in long-read single-cell RNA sequencing

Abstract

Differential isoform usage has been reportedly associated with distinct contexts associated with cell-type and disease; however, the mechanistic role that individual isoforms play in these contexts is poorly understood. Long-read single-cell RNA sequencing enables greater resolution into genes and isoforms previously difficult to detect via short-read approaches, including pseudogenes and those with repetitive motifs. This technique, combined with isoform-discovery tools such as Bambu, has yielded thousands of newly discovered isoforms and genes for consideration. The functional relevance of these new isoforms, however, remains enigmatic. Tools to prioritize which of these novel isoforms should be interrogated and to predict their functionality remain elusive. To address this, we developed a computational framework, Centroid-based Isoform Pathway Inference (CIPI), which contextualizes novel genes and isoforms relative to reference genes with similar cell-type expression patterns. Rather than relying solely on host-gene annotation or similarity to known protein domains, CIPI constructs expression-based neighborhoods that, combined with pathway analyses such as GO, KEGG, or Reactome, enable inference of pathway-level contextualization of individual isoforms according to their biologically meaningful neighbors. Combining CIPI with pathway analyses provides a biological context to novel isoforms, enabling a hypothesis-generating approach for prioritizing novel isoforms across the isonome.