For a better presentation of this page, you can also visit Gitbook.
We have developed the NiCo (Niche Covariation) package for the integration of single-cell resolution spatial transcriptomics and scRNA-seq data. NiCo allows you to:
(1) Perform cell type annotations in the spatial modality by label transfer.
(2) Predict niche cell type interactions within local niche neighborhoods.
(3) Infer cell state covariation and the underlying molecular crosstalk in the niche.
NiCo infers factors capturing cell state variability in both modalities and identifies genes correlated to these latent factors to predict ligand-receptor interactions and factor-associated pathways.
NiCo can be used on data generated by any spatial transcriptomics method, but it is mainly designed for single-cell resolution spatial technologies such as MERFISH, seqFISH, Xenium etc.
We are providing a first tutorial on running the NiCo pipeline for the data integration of single-cell RNA-sequencing (reference) and single-cell resolution spatial transcriptomics data (query).
The second NiCo tutorial showcases an application to low-resoution genome-wide sequencing-based spatial transcriptomics data where integration with single-cell RNA-sequencing data is not required.
Please use the tutorial link below and keep all the files (NiCoLRdb.txt and *.ipynb) and folders (inputRef, inputQuery; after unzipping) in the same path to complete the tutorial.
- Part A: Data Preparation Extract all zip files and run the Juypter notebook
Start_Data_prep.ipynb
to create the data files for NiCo analysis. - Part B: Running NiCo on selected cell types After data preparation and generating normalised data files, run the Jupyter notebook
nico_analysis_highres_image_tech.ipynb
to perform the core steps of NiCo.
This tutorial explains all steps of the NiCo pipeline, including annotation of cell types in the spatial modality by label transfer from the scRNA-seq data, prediction of significant cell type niche interactions, and derivation of cell state covariation within the local niche.
By default, the figures generated are saved both in the respective directory and inside the notebook.
Please refer to the documentation for details on functions and parameters.
The data source is provided in the manuscript. - Part B: Running NiCo on all cell types If users want to perform NiCo analysis on the full data without specifying any cell type, refer to the script.
nico_analysis_highres_image_tech.py
. The output log can be seen here. Due to large number of images, a Jupyter notebook might not display them properly. However, leaving thechoose_celltypes
andchoose_factors_id
lists blank will enable the full analysis.
Run NiCo on cerebellum data from Slide-seqV2 technology [data from Cable, D. M. et al. Nature methods 19, 1076–1087 (2022)].
- Download the data from this link and place the data in the following path to complete the tutorial:
nico_cerebellum/cerebellum.h5ad
- The NiCo niche detection and covariation analysis tasks can be run via following the jupyter notebook
nico_analysis_lowres_seq_tech.ipynb
Please follow the instructions provided in the nico-sc-sp pip repository for set up and installation.
For detailed instruction, visit the nico-sc-sp pip repository
conda create -n nicoUser python=3.11
conda activate nicoUser
pip install nico-sc-sp
Sometimes, the pygraphviz package cannot be installed via pip, or during the cell type interaction part it returns an error that "neato" is not found in the path, so an alternative installation via conda is recommended. Please follow the installation of pygraphviz here
conda create -y -n nicoUser python=3.11
conda activate nicoUser
conda install -c conda-forge pygraphviz
pip install nico-sc-sp
pip install jupyterlab
To check the usage of any function you can execute the command as in the example:.
[<moduleName>.<functionName>.__doc__
]
print(scov.save_LR_interactions_in_excelsheet_and_regression_summary_in_textfile_for_interacting_cell_types.__doc__)
If at any step returns the warning findfont: Font family 'Helvetica' not found
please initialize the matplotlibrc file to use a different font:
import matplotlib.pyplot as plt
plt.rcParams['font.family'] = 'sans-serif'
plt.rcParams['font.sans-serif'] = ['Tahoma', 'DejaVu Sans','Lucida Grande', 'Verdana']
If you are working with Vizgen MERSCOPE spatial data, please process the output with the "process_vizgenData.py" script to convert Vizgen data into gene_by_cell.csv and tissue_positions_list.csv files.
If you encounter any issues while running, please ensure that the cell barcode name is composed of characters rather than integer numbers. When pandas reads numeric values, it will read as int64 instead of objects which creates datatype confusion in other parts of the code. Therefore, please convert your cell barcode numbers to strings if they are purely numeric.
For detailed documentation of NiCo modules and their usage, visit NiCo documentation.
If you use NiCo in your research, please cite it as follows:
- Ankit Agrawal, Stefan Thomann, Sukanya Basu, Dominic Grün. NiCo Identifies Extrinsic Drivers of Cell State Modulation by Niche Covariation Analysis. Submitted, 2024
contact: If you face any problem during the tutorial or have any questions, please email me (ankitplusplus at gmail.com) or raise an issue on GitHub.