HumanMorphoMap: Multimodal Data Fusion of Human Cortical Neurons

HumanMorphoMap is the official code repository for the study:
"Multimodal Data Fusion Reveals Morpho-Genetic Variations in Human Cortical Neurons Associated with Tumor Infiltration"

This framework integrates high-resolution 3D neuronal morphology (reconstructed via Let'sACT) with spatial (10x Visium) and bulk transcriptomics to quantify how glioma infiltration reshapes human cortical neurons.

---

🚀 Key Features

• Morphological Profiling: Automated reconstruction of 3D morphologies of 8,398 human cortical neurons covering various regions.
• Tumor vs. Normal Comparison: Statistical pipelines to compare infiltrated tissues vs. normal tissues.
• Transcriptomic Integration: Tools to map gene expression gradients (e.g., CDKN2A, TP53) to morphological phenotypes.
• Spatial Mapping: Correlating morphological atrophy with tumor infiltration using spatial transcriptomics spots.

---

🛠 Installation

Prerequisites

• Linux. The source code and dependencies are cross-platform. While currently tested only on Linux, they are expected to work on macOS and Windows.
• Python ≥ 3.9.

Key Dependencies

• Vaa3D (Visualization and feature calculation. Pre-built Vaa3D-x version 1.1.4)
• pylib (Customized Python library for neuron image/morphology processing)
• scanpy (Transcriptomic analyses. Tested on version 1.11.2)
• cell2location (Spatial deconvolution. Tested on version 0.1.4)
• numpy, pandas, scipy (Core computation. Tested on version 1.26.4, 2.1.1, and 1.11.2 respectively)
• matplotlib, seaborn (Visualization. Tested on version 3.6.0, 0.13.0)

Setup Environment

We recommend using conda to manage dependencies.

# Clone the repository
git clone [https://github.com/SEU-ALLEN-codebase/HumanMorphoMap.git](https://github.com/SEU-ALLEN-codebase/HumanMorphoMap.git)
cd HumanMorphoMap

# Create a virtual environment
conda create -n human_morpho python=3.10
conda activate human_morpho

Install dependencies

1. Internal Library (pylib) This project depends on our internal library, pylib. Please clone the repository and add its location to your PYTHONPATH:

git clone https://github.com/SEU-ALLEN-codebase/pylib.git
export PYTHONPATH=$PYTHONPATH:/path/to/pylib

2. Standard Dependencies Install the remaining packages via pip. The installation of dependencies should be within minutes, and it was tested on Ubuntu 20.04 and 24.04.

User Guide

The overall structure of the project:

HumanMorphoMap/
├── meta/                       # Meta processing
├── src/                        # Analytical or visualization utilities. 
├── common_utils/          
├── h01-guided-reconstruction/  # Utilities for EM-based reconstruction optimization
├── human_glioma_CGGA/src/      # Bulk transcriptomics analyses
├── resources/                  # Utilities for processing publicly downloaded morphological datasets
├── soma_morphology/            # Evaluation of soma morphology
├── soma_normalized/            # Post-processing the reconstructons and their features 
├── spatial-enhanced/           # Deprecated
├── spatial_transcript_seu      # Utilities for spatial transcriptomic data processing
├── LICENSE
└── README.md

The source code is organized by function and corresponds directly to the figures in the manuscript. To execute an analysis, simply update the file paths under if __name__ == '__main__' in the relevant script.

📂 Data Availability

• The datasets generated in this study, including automated and manual neuronal reconstructions (.swc format) and spatial transcriptomics data, have been deposited on Zenodo (DOI: 10.5281/zenodo.15189542). Comprehensive metadata is available within the repository and in the Supplementary Information accompanying this manuscript.
• Bulk transcriptomic data are downloaded from Chinese Glioma Genome Atlas (CGGA) via https://www.cgga.org.cn/download.jsp.

📜 Citation

If you use this code or data in your research, please cite our preprint:

@article{Liu2025HumanMorphoMap,
  title={Multimodal Data Fusion Reveals Morpho-Genetic Variations in Human Cortical Neurons Associated with Tumor Infiltration},
  author={Yufeng Liu, Zhixi Yun, et al.},
  journal={bioRxiv},
  year={2025},
  doi={10.64898/2025.12.26.696632v2},
  url={[https://www.biorxiv.org/content/10.64898/2025.12.26.696632v2](https://www.biorxiv.org/content/10.64898/2025.12.26.696632v2)}
}