Quantitative AI-based DNA fiber workflow to study replication stress

Abstract

Replication stress (RS) is a prominent source of genome instability and human diseases. Understanding its molecular mechanism through various quantitative and unbiased methodologies is essential for the advancement of treatment strategies. One of the powerful methods to study DNA replication dynamics and its alterations at the single-molecule resolution is the DNA fiber assay. However, this method relies exclusively on manual image acquisition and analysis, making it time-consuming and susceptible to user bias. Here, we present a quantitative AI-based DNA fiber (qAID) workflow enabling imaging and multiparameter analysis of thousands of DNA fibers within several dozen minutes. Our workflow quantifies key parameters, including DNA fiber frequency, length, and symmetry, while also allowing visual inspection of individual DNA fibers using unbiased image galleries. The robustness of the workflow is demonstrated by comprehensive datasets of biologically relevant experiments performed by three independent laboratories. Overall, qAID workflow provides a fast and effective examination of replication dynamics and its alterations at the single-molecule resolution.