Researchers have introduced MaGNet, an open-source analytical tool that applies network theory to evaluate and quantify the branching architecture of the mammary ductal tree in developing female mice. Reported in the Journal of Mammary Gland Biology and Neoplasia, the computational method establishes a quantitative framework for analyzing whole-mount mammary tissue preparations across pubertal stages and hormonal interventions.
The mammary gland represents an unusually dynamic organ system with a complex branching architecture that develops entirely after birth in response to circulating hormonal cues. In mammary gland biology, characterizing branching morphogenesis serves as a primary experimental strategy to determine how developmental, physiological, and molecular perturbations influence tissue invasion, ductal expansion, and structural maintenance. Despite widespread reliance on whole-mount tissue imaging across experimental laboratories, the discipline has historically lacked a standardized, fully open-sourced computational framework to extract reproducible quantitative measurements from these preparations.
To address this technical gap, the research team engineered the Mammary Gland Network analysis tool, or MaGNet, which translates physical ductal configurations into formal mathematical network representations. By applying this automated pipeline to brightfield micrographs of murine mammary glands collected at three distinct pubertal timepoints, the investigators demonstrated reproducible measurement of ductal network growth across progressive stages of normal maturation.
Beyond baseline pubertal maturation, the analytical pipeline captured ductal expansion patterns elicited by exposure to pregnancy-related hormones. The researchers observed that hormonal stimulation triggered measurable structural remodeling in the mammary ductal architecture, indicating that network-based topological metrics remain sensitive to hormone-induced tissue expansion in adult female murine models.
The experimental validation of the pipeline utilized defined cohorts of laboratory mice maintained in compliance with institutional animal care guidelines approved by the Cold Spring Harbor Laboratory Institutional Animal Care and Use Committee. For pubertal time-course evaluations, wildtype female C57BL/6 mice were assessed at ages of one month, one and a half months, and two months, sourced from Charles River. For parallel hormone exposure assessments, the investigators evaluated six- to eight-week-old female Balb/C mice subjected to controlled hormonal regimens.
To prepare tissue for structural quantification, the lower left mammary gland was dissected from each animal and spread flat onto a glass microscope slide according to established whole-mount staining protocols. The harvested tissue was fixed overnight at four degrees Celsius in Carnoy’s fixative, composed of three parts absolute ethanol to one part glacial acetic acid, after which the fixative was rinsed away and the samples were cleared in graded alcohol and tap water. The glands were subsequently incubated in Alum carmine staining solution for at least twenty-four hours at four degrees Celsius, subjected to serial ethanol dehydration washes at ascending concentrations, and cleared in histoclear solution until the surrounding mammary fat pad became transparent.
Stained whole-mount preparations were immersed in histoclear mounting medium under glass coverslips and photographed on a Nikon SMZ25 brightfield microscope equipped with a Nikon DS-Ri2 stereoscope camera. The resulting high-resolution brightfield images were manually traced on an iPad using an Apple Pencil within the Sketch application, isolating the ductal tree onto an independent graphic layer over a uniform white background. To eliminate individual operator bias during manual contouring, tracings were performed independently by multiple investigators before each layer was exported as an uncompressed TIF image.
The exported TIF tracings were imported into the Network Extraction From Images software, known as NEFI, to convert manual drawing contours into rigorous mathematical network structures. Using a predefined leaf venation extraction workflow, NEFI parsed the traced ductal geometries and outputted spatial coordinate data representing the nodes and interconnecting segments. The pipeline then executed automated filtering on the resulting spatial data to discard disconnected fragments and isolate the primary connected component of the ductal tree.
The coordinate matrices generated by NEFI were ingested into a custom Python workflow constructed around the NetworkX network analysis package, using an internal read_nefi_graph function that converts spatial coordinates into native NetworkX graph structures. The pipeline allows rapid visual inspection through a quick_plot function while computing a comprehensive suite of topological parameters. These parameters include total edge count corresponding to ductal segments, terminal node counts representing terminal end buds, individual and cumulative edge lengths, total branching tree length normalized by tissue area, the convex hull area enclosing the ductal network, and the distribution of branching degrees across the gland.
All downstream statistical comparisons and visual figure generations were executed in Python, utilizing Tukey’s honestly significant difference tests implemented through the pairwise_tukeyhsd function in the statsmodels package alongside custom plotting functions developed with Seaborn. To promote reproducibility across diverse developmental stages and related branching organ models, the authors released the complete computational pipeline as an interactive Google Colab notebook alongside demonstration datasets hosted in a public GitHub repository.
In evaluating the biological responses of murine tissue, the experimental design distinguished between pubertal progression and hormonal stimulation through timed animal cohorts. For the pubertal series, wildtype female C57BL/6 mice were obtained from Charles River at one month, one and a half months, and two months of age to evaluate ductal branching during development. For hormone exposure trials, investigators obtained six- to eight-week-old female Balb/C mice from Charles River. The structural impact of pregnancy hormones was evaluated across distinct durations, analyzing three whole-mount images for the six-day exposure condition alongside four whole-mount images for the twelve-day exposure condition. All murine procedures were completed in agreement and approval with the Cold Spring Harbor Laboratory Institutional Animal Care and Use Committee.
The histological preparation followed a chemical protocol to preserve the structural features of the ductal network before microscopic evaluation. After dissection of the lower left mammary gland, the tissue was spread carefully on a glass slide. The gland was fixed overnight at four degrees Celsius with Carnoy’s fixative, prepared with three parts ethanol to one part glacial acetic acid. Following fixation, the Carnoy’s fixative was discarded, and the tissue was incubated in seventy percent ethanol for fifteen minutes, followed by a wash with running tap water for five minutes. Whole-mount staining was conducted with filtered Alum carmine solution, containing one gram of carmine dye and two and a half grams of potassium alum in five hundred milliliters of MilliQ water. The tissue was incubated in this staining solution for at least twenty-four hours at four degrees Celsius, followed by serial fifteen-minute washes in ethanol solutions of seventy percent, ninety-five percent, and absolute concentration, with each wash repeated once.
Following dehydration, the stained tissue was immersed in histoclear solution overnight or until the fat pad cleared completely before imaging. Slides were covered with coverslips using the same histoclear media as the mounting medium during image acquisition. Micrographs were acquired on a Nikon SMZ25 brightfield microscope equipped with a Nikon DS-Ri2 stereoscope. The authors noted that for long-term slide storage, whole-mount mammary tissue slides can be mounted with permount and stored at room temperature shielded from light. To extract the branching network from brightfield images, manual tracing was performed on an iPad using an Apple Pencil within the Sketch application. The tracing was saved as an independent layer on a white background. Tracings were completed by multiple individuals to remove potential biases in individual tracings, after which each layer was converted separately into a TIF file.
Network Extraction From Images, or NEFI, was used to go from manual tracings to mathematical representations of network structure. The manual tracings entered the software as TIF files, and the pre-defined NEFI pipeline Leaf Venation was used to convert each tracing into a network object. Filtering was performed on the main mammary gland object to remove all connected components not belonging to the main connected component. This step ensured that only the primary connected structure was retained for quantitative analysis. Once extraction and filtering were completed, NEFI outputted spatial network coordinates as an output.txt text file, providing the coordinates required for graph analysis.
To analyze and quantify mammary gland network features, the researchers developed a Python pipeline using the NetworkX python package. Spatial coordinates from the NEFI output.txt file were transformed into a NetworkX Python object using a defined read_nefi_graph function. The resulting graph can be visualized using a quick_plot function. The pipeline supports calculation of network properties, including the number of edges representing ducts, the number of terminal nodes representing terminal end buds, edge lengths, normalized length of the branching tree by tree area, convex hull area occupied by the ductal network, and degree branching distribution. All statistical analyses and data visualizations from raw data were conducted in Python. Statistical comparisons between groups used Tukey’s honestly significant difference tests performed with the pairwise_tukeyhsd function in the statsmodels package, while data visualizations were developed using the Seaborn package.
Addressing data availability, the authors noted that no datasets were generated or analysed during the study. Code for the pipeline was established as an interactive notebook openly available on Google Colab under the title Mammary gland analysis on NetworkX. Demonstration data folders were also uploaded to GitHub at https://github.com/lucia-tellez/NetworkAnalysis/tree/main for users to test the functionalities. By distributing open-source tools to the research community, this computational methodology was designed to increase reproducibility and broad applicability across diverse organ systems, model organisms, and developmental stages where branching morphogenesis is evaluated.
Subject of Research: Network-based quantitative analysis of mouse mammary gland branching morphogenesis
Article Title: MaGNet: A Network-Based Method for Quantitative Analysis of the Mammary Ductal Tree in Developing Female Mice
Article References: Lewis, S. M., Tellez-Perez, L., Henry, S., Zheng, X., Navlakha, S., & dos Santos, C. O. (2025). MaGNet: A Network-Based Method for Quantitative Analysis of the Mammary Ductal Tree in Developing Female Mice. Journal of Mammary Gland Biology and Neoplasia, 30(1), Article 13. https://doi.org/10.1007/s10911-025-09589-1
Image Credits: AI Generated
DOI: 10.1007/s10911-025-09589-1
Keywords: mammary gland biology, branching morphogenesis, MaGNet pipeline, network theory, ductal tree expansion, whole-mount imaging, C57BL/6 mice, Balb/C mice, pubertal development, pregnancy hormones, Alum carmine staining, Network Extraction From Images, NetworkX, terminal end buds, convex hull area, Tukey HSD test, Python data visualization
Cite Scienmag News
Drew Townsend. (November 22, 2026). Open-Source Network Tool Quantifies Mammary Ductal Branching in Developing Mice. Scienmag. https://scienmag.com/open-source-network-tool-quantifies-mammary-ductal-branching-in-developing-mice/
Drew Townsend. "Open-Source Network Tool Quantifies Mammary Ductal Branching in Developing Mice." Scienmag, 22 November 2026, https://scienmag.com/open-source-network-tool-quantifies-mammary-ductal-branching-in-developing-mice/. Accessed 22 November 2026.
Drew Townsend. "Open-Source Network Tool Quantifies Mammary Ductal Branching in Developing Mice." Scienmag. November 22, 2026. https://scienmag.com/open-source-network-tool-quantifies-mammary-ductal-branching-in-developing-mice/








