Cellular morphology emerges from polygenic, distributed transcriptional variation

Height and most disease risk are known polygenic traits: characteristics governed by multiple genes at different loci instead of a select few. Though we are beginning to understand how genetic variation impacts cell morphology, whether such an analogous polygenic architecture operates at the cellular level, where morphology integrates cytoskeletal organization, organelle positioning, and metabolic state, has yet to be systematically tested. Here, we demonstrate that cellular morphology behaves as a polygenic trait by integrating multimodal modeling, perturbation profiling, and population scale genetic variation. A shared latent-space autoencoder trained on four large perturbation datasets predicts morphology from gene expression and generalizes without retraining to matched RNA-seq and Cell Painting profiles from 100 genetically diverse iPSC donors. The model predicted 17 morphological features (R > 0.6, permutation FDR q < 0.05), enriched for spatial organelle distribution and cytoskeletal architecture. Predictive performance does not arise from dominant gene-phenotype relationships: individual genes contribute modestly, and marginal gene-morphology correlations are uniformly weak, revealing a distributed regulatory architecture. Despite this polygenicity, CRISPR perturbation data from the JUMP consortium validates specific model-prioritized genes, such as the cytoskeletal regulator TIAM1, membrane trafficking factor RAB31, and mitochondrial-associated membrane transporter ABCC5, as molecular anchors whose disruption produces feature-specific morphological shifts. Transcriptome-wide association analyses identify correlational variant-gene-morphology chains linking cis-regulatory variation through mitochondrial metabolism (PDHX) and iron transport (SLC11A2) to cellular architecture. These results establish cellular morphology as a polygenic systems phenotype, extending the omnigenic framework to the cellular level and providing a biological basis for interpreting cross-modal prediction in functional genomics. ### Competing Interest Statement The authors have declared no competing interest. AnalytiXIN Fellowship in Life Sciences

Spring Science

Giving Scientists AI Superpowers

Data Scientist

Form

CellDeathPred: a deep learning framework for ferroptosis and apoptosis prediction based on cell painting - Cell Death Discovery

Cell death, such as apoptosis and ferroptosis, play essential roles in the process of development, homeostasis, and pathogenesis of acute and chronic diseases. The increasing number of studies investigating cell death types in various diseases, particularly cancer and degenerative diseases, has raised hopes for their modulation in disease therapies. However, identifying the presence of a particular cell death type is not an obvious task, as it requires computationally intensive work and costly experimental assays. To address this challenge, we present CellDeathPred, a novel deep-learning framework that uses high-content imaging based on cell painting to distinguish cells undergoing ferroptosis or apoptosis from healthy cells. In particular, we incorporate a deep neural network that effectively embeds microscopic images into a representative and discriminative latent space, classifies the learned embedding into cell death modalities, and optimizes the whole learning using the supervised contrastive loss function. We assessed the efficacy of the proposed framework using cell painting microscopy data sets from human HT-1080 cells, where multiple inducers of ferroptosis and apoptosis were used to trigger cell death. Our model confidently separates ferroptotic and apoptotic cells from healthy controls, with an average accuracy of 95% on non-confocal data sets, supporting the capacity of the CellDeathPred framework for cell death discovery.