Imaging-based CRISPR screens reveal gene functions through cellular morphology, advancing insights into pathways and diseases
Over the past decade, CRISPR technology—a groundbreaking tool that allows scientists to precisely edit genes within living cells—has transformed biology by enabling unprecedented control over genetic information. While traditional methods relied heavily on measuring simple outcomes such as cell survival, today’s sophisticated image-based analyses vividly illustrate how gene edits reshape the very appearance of cells.
This shift from traditional, binary CRISPR screens—focusing mainly on endpoints like cell survival—to advanced image-based analyses marks a leap akin to moving from snapshots to streaming video. Researchers today examine nuanced cellular responses, such as shifts in organelle distribution, alterations in cell shape, or changes in nuclear organization. These visual changes collectively termed “cell morphology,” provide richer insights into gene function.
In 2019, researchers used CRISPR in combination with fluorescent imaging to study NF-κB signaling—a critical cellular pathway controlling immune responses. By observing protein dynamics rather than static endpoints, Feldman and colleagues at Broad institute – a world leading research institute in biology – identified unexpected roles for genes MED12 and MED24. Such real-time insights highlight the importance of timing and spatial context in understanding gene function.
This approach reached new heights in a landmark 2025 study by Ramezani et al., who performed genome-wide CRISPR screenings paired with a technique called “Cell Painting”—an imaging assay that labels various organelles and cellular compartments. Their extensive “cell morphology atlas” linked gene knockouts with specific visual outcomes, uncovering surprising roles for obscure genes like TMEM251, which emerged as critical for lysosomal trafficking. Such discoveries could have direct implications for understanding lysosomal storage disorders—diseases caused by defects in how cells handle waste.
Other recent studies reinforce this approach’s power: Wong et al. (2024) identified numerous genes essential for moving organelles inside cells, while Wang et al. (2023) revealed how disrupting genes involved in DNA packaging profoundly changes nuclear structure. Both discoveries show that visual changes in cell shape often mirror deeper functional disruptions.
Deepcell’s recent work in collaboration with scientists at EMBL, a leading molecular biology research organization in Heidelberg, Germany presented at a ASHG conference amplifies these developments further. By harnessing deep learning algorithms to analyze cellular images, Deepcell’s platform detected subtle yet biologically meaningful differences between CRISPR knockout cell lines (see Fig. 1 below). Unlike traditional methods that depend on human interpretation, these neural networks automatically uncover detailed morphological patterns linked to specific genetic edits.
Figure 1: Density plot of UMAPs projections generated by Deepcell REM-I shows morphological shifts consistent with biochemical and molecular functions.
Moreover, Deepcell’s morphological profiling, driven by machine learning, significantly improves the efficiency and cost-effectiveness of functional genomics. Traditional gene-by-gene analyses are laborious and expensive; by contrast, Deepcell looks at the entire cell, dramatically increasing throughput and reducing costs. This technological advancement democratizes high-dimensional phenotyping, making it accessible to a broader range of laboratories.
Looking forward, the integration of CRISPR, high-dimensional imaging, and artificial intelligence promises to redefine precision medicine and drug discovery. Rapidly visualizing how genetic mutations affect cell behavior could speed up the identification of disease-causing variants and effective treatments, reshaping medicine as we know it. Moreover, genes with similar functions are clustering based on morphology—for example, VAC14 and FIG4, both of which regulate vacuole function, exhibit shared morphological traits. This suggests that morphological signatures could serve as a powerful tool for inferring gene function, where similar morphologies indicate similar biological roles.
In biology, appearances, it seems, are far from superficial. CRISPR-based morphological profiling is not merely an incremental advance—it represents a fundamental paradigm shift. Through the powerful lens of imaging, we are finally capturing the full complexity of how genes shape life at the cellular level.
Mahyar Salek
CTO
