spaCR: Spatial phenotype analysis of CRISPR-Cas9 screens

Pooled CRISPR-Cas9 screens have emerged as powerful functional genomic tools. However, they typically rely on aggregate-level or transcriptomic outputs, which limits their ability to capture spatially resolved phenotypes. We developed spatial phenotype analysis of CRISPR-Cas9 screens (spaCR), an open-source software package that integrates deep-learning single-cell image classification with barcode-based, well-level genotype mapping to link pooled genetic perturbations to spatial phenotypes. Unlike optical pooled screening or in situ genotyping methods, spaCR requires only standard imaging and sequencing platforms, eliminating the need for specialized equipment. To guide experimental design, we developed spaCRPower, a statistical simulation tool that models screen parameters and estimates power for detecting genotype–phenotype associations. As a proof-of-concept, we applied spaCR to a pooled Toxoplasma mutant library to identify parasite effectors involved in hijacking the host ESCRT machinery. We identified both an established and a previously uncharacterized genetic determinant, EAF1, highlighting spaCR as a powerful tool for mapping genotype–phenotype relationships.