Deep Learning of Functional Perturbations from Condensate Morphology

Biomolecular condensates compartmentalize the interior of living cells to spatiotemporally organize complex functions, yet linking molecular interactions within condensates to their mesoscale organization remains a major challenge. To bridge this gap, we developed a neural network-based framework - Deep-Phase - that uses microscopy images to quantitatively classify condensate morphology changes resulting from pharmacological alterations in associated biochemical processes. We use Deep-Phase to precisely quantify time- and concentration-dependent structural perturbations to the multiphase nucleolus and show that they are tightly coupled to potencies of drugs inhibiting rRNA transcription and processing. Applying Deep-Phase in a chemical screen, we identify a unique nucleolar morphology and discover a role for a DNA topoisomerase in rRNA processing. Mechanistic studies of this morphology provide insights into how interfaces between nucleolar subcompartments are maintained. We demonstrate Deep-Phase’s adaptability to diverse cell lines, labels, and condensates, offering a powerful platform for uncovering cellular organizing principles and therapeutic targets. A deep learning framework, Deep-Phase, classifies and quantifies drug-induced changes in morphologies of nucleoli, nucleolar speckles, and viral cytoplasmic condensates, directly from images. Time- and concentration-dependent morphological responses to perturbation predict associated disruptions in RNA transcription and processing. Using Deep-Phase in a high-content small molecule screen reveals a unique nucleolar morphology induced by TOP1 inhibition. TOP1 inhibition leads to reduced levels and processing of large ribosomal subunit precursors and provides a mechanism for maintenance of nucleolar phase boundaries.