Short-read and long-read single-cell sequencing capture distinct perturbation effects in CRISPR screens

Background: Single-cell CRISPR screens have transformed functional genomics by enabling scalable, systematic investigation of gene function. Despite this, transcriptional complexity has largely been overlooked, with studies focusing on gene-level effects rather than isoforms. Methods capable of capturing splicing and isoform usage have emerged, including long-read sequencing and alternative library preparation strategies, but their suitability for large-scale perturbation screens remains unevaluated. Results: We compare two single-cell library preparation methods (5' 10x Genomics and Parse Biosciences Evercode) across Illumina short-read, Oxford Nanopore Technologies, and PacBio long-read sequencing, applying CRISPRi to silence three genes with distinct regulatory roles (DDX6, GEMIN5 and GFI1B) in K562 cells. While short-read methods can detect some alternative splicing events, only long-read sequencing consistently captures isoform-level changes. Although Parse Evercode provided even coverage across transcripts, we observed strong enrichment of intronic reads, limiting its utility for splicing analysis. The primary constraint of long-read approaches was depth: approximately 21 million reads are required for 80% saturation of splicing events in a single perturbation, underscoring the need for higher-throughput methods. Despite these limitations, we show that GEMIN5 perturbation produced modest differential expression but the most extensive splicing changes, an effect invisible to gene-level analysis, highlighting the importance of extending CRISPR screens to isoform-level readouts. Conclusions: We provide a practical framework for isoform-level analysis in single-cell CRISPR screens, identifying both the capabilities and current limitations of available approaches. As perturbation studies scale, long-read sequencing will be essential for comprehensive functional interpretation, capturing biology missed by gene-level analysis.