BACKGROUND: Quantitatively tracking subcellular protein puncta over time is critical for understanding protein functions and related cellular processes, as many proteins assemble into discrete protein puncta when performing functions. Monitoring these protein puncta over an extended period enables the quantitation of their dynamic behaviors, such as repositioning, remodeling, and inter-object interactions such as fusion and split. These characteristics are essential for deciphering the underlying regulatory mechanisms of protein function. However, tracking protein puncta is challenging due to these clusters undergoing rapid and complex temporal changes. RESULTS: To address these challenges, a comprehensive tool ‘ProTrack3D’ is developed. It implements a multi-stage multi-object tracking workflow specifically designed to handle the complexity of dynamic protein puncta. The segmentation stage can adopt any advanced deep neural network to detect protein puncta. The Tracking stage follows individual puncta over time and detects birth, death, split and fusion events. The Life Path Reconstruction stage visualizes the life history of protein puncta using tree structures. Applying ProTrack3D tool to the 4D fluorescence microscopy images of developing Drosophila embryos confirms the effectiveness of the method implemented at each stage. Compared with existing tracking tools, our tracking algorithm achieves significant improvement in tracking performance by utilizing deep feature maps which encode rich spatial and intensity information. CONCLUSIONS: By integrating all stages of tracking and offering quantitative analysis functions with a user-friendly graphical interface, ProTrack3D enables researchers with basic computer skills to perform segmentation, tracking, and analysis of protein puncta in fluorescence microscopy images, thus facilitating the broader accessibility and usability of advanced protein tracking techniques. ProTrack3D is available for distribution via GitHub at https://github.com/ramugautam1/ProTrack3D .

MOTIVATION: While multi-channel fluorescence microscopy is a vital imaging method in biological studies, the number of channels that can be imaged simultaneously is limited by technical and hardware limitations such as emission spectra cross-talk. One solution is using deep neural networks to model the localization relationship between two proteins so that the localization of one protein can be digitally predicted. Furthermore, the input and predicted localization implicitly reflect the modeled relationship. Accordingly, observing the response of the prediction via manipulating input localization could provide an informative way to analyze the modeled relationships between the input and the predicted proteins. RESULTS: We propose a protein localization prediction (PLP) method using a cGAN named 4D Reslicing Generative Adversarial Network (4DR-GAN) to digitally generate additional channels. 4DR-GAN models the joint probability distribution of input and output proteins by simultaneously incorporating the protein localization signals in four dimensions including space and time. Because protein localization often correlates with protein activation state, based on accurate PLP, we further propose two novel tools: digital activation (DA) and digital inactivation (DI) to digitally activate and inactivate a protein, in order to observing the response of the predicted protein localization. Compared with genetic approaches, these tools allow precise spatial and temporal control. A comprehensive experiment on six pairs of proteins shows that 4DR-GAN achieves higher-quality PLP than Pix2Pix, and the DA and DI responses are consistent with the known protein functions. The proposed PLP method helps simultaneously visualize additional proteins, and the developed DA and DI tools provide guidance to study localization-based protein functions. AVAILABILITY AND IMPLEMENTATION: The open-source code is available at https://github.com/YangJiaoUSA/4DR-GAN. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The in-vivo functions of many proteins highly depend on their ability to assemble into supramolecular complexes at specific subcellular localizations. Such subcellular localizations can change in response to intracellular and extracellular cues, which is an important way to regulate the behavior and morphology of the cell. Therefore, segmentation and tracking of these supramolecular complexes are critical in understanding gene functions and regulations. Proteins that form discrete puncta or clusters can be tracked for a long period of time. However, segmentation and tracking become challenging when protein complexes undergo quick repositioning and remodeling such as assembly, disassembly, fusion, and split. In this paper, we first use deep learning methods to segment and track protein clusters with changing morphology, size, and intensity in hyperdimensional biological images. Using the segmentation and tracking results, we reconstruct the life paths and family tree of the protein clusters by integrating fusion and split events. Based on the family trees, various quantitative analyses can be performed including temporal correlations between the tracked protein and its interacting protein complexes. The effectiveness of the proposed method for analyzing subcellular proteins is confirmed through the evaluation using the two-channel 3D fluorescent microscopy time-lapse videos (5D images) of developing Drosophila embryos.