Mickelsen et al 2020 scRNA-seq of the Ventroposterior Hypothalamus

This repository accompanies the preprint L.E. Mickelsen, W.F. Flynn, et al..

Laura E Mickelsen, William F Flynn, Kristen Springer, Lydia Wilson, Eric J Beltrami, Mohan Bolisetty, Paul Robson, Alexander C Jackson. "Cellular taxonomy and spatial organization of the murine ventral posterior hypothalamus." eLife (2020)

Current Status

• Manuscript: public
• Raw Data: public
• Figure code: public (here!)
• Analysis code: public-ish (here!).
• Finalized h5ad files: public
• Finalized Loupe Cell Browser files: public
• Marker genes for every cluster and subcluster: public

Note on analysis code: The original analysis code makes heavy use of a custom wrapper around Scanpy 1.3.7 that can be found here. This version of the code is largely readable but not functional unless you install scanpy_recipes. In order to make this work more accessible, I am porting the analysis code to Scanpy 1.4.5+ but this is a work in progress.

Sample descriptions

Ventral posterior hypothalamus microdissections were harvested from four (4) batches of three (3) mice. Two batches were harvested at a time, one male batch and one female batch:

LibraryID	Batch	Strain	Age	Sex	Harvest	Time of Harvest	Chemistry Version
AJ18003	1	C57BL/6	P30-35	Male	1	March 2018	v2
AJ18004	2	C57BL/6	P30-35	Female	1	March 2018	v2
AJ19001	3	C57BL/6	P30-35	Female	2	March 2019	v3
AJ19002	4	C57BL/6	P30-35	Male	2	March 2019	v3

Combined data

You can interactively analyze the resulting aggregated data using the 10X Loupe Browser by downloading .cloupe files we've prepared using our datasets. Those data will be available for download here:

https://singlecell.jax.org/hypothalamus

We have provided 3 Loupe files for the aggregated global clustering, neuronal sub-clustering, and non-neuronal subclustering.

When using the Loupe files, note the following:

• As of March 2020, all Loupe files will contain default t-SNE and UMAP embeddings as well as default K-means and Graph-based clustering that was generated by 10X's cellranger tool. It is currently impossible to remove these entries from the Loupe file.

• To circumvent these, we have added our UMAP embedding under the name "CustomUMAP" and our clustering under the name "Custom Clustering". Please refer to these when trying to correlate the Loupe file with results presented in the manuscript.

Overview of the analysis

Analysis code

Most of the original analysis was done with Scanpy v1.3.7. However, I have updated the majority of the analysis to be compatible with Scanpy v1.4+. Currently finishing that effort up now.

Reproducing manuscript figures

The manuscript figures were generated as individual panels which were then combined in Adobe Illustrator. However, the individual panels showing scRNA-seq can be regenerated using notebooks in notebooks/ and associated utility scripts. You can view those directly here:

• Main figures
• SI figures
• Extra figures

Requirements

Full requirements are listed in the environment.yaml file, but for the most part, the following Python 3.6+ packages are needed.

scanpy
numpy
pandas
scikit-learn
matplotlib
seaborn
cmocean

Downloading data

Some of the data is too large to host on GitHub. Download the data directly via your web browser (ftp://ftp.jax.org/flynnb/mickelsen-vph-2020/) or use the command line:

git clone [email protected]:TheJacksonLaboratory/ventroposterior-hypothalamus-scrna-seq.git
cd ventroposterior-hypothalamus-scrna-seq/data
wget ftp://ftp.jax.org/flynnb/mickelsen-vph-2020/h5ad.zip
unzip h5ad.zip

Running the notebooks

jupyter notebook notebooks/main_figures.ipynb
jupyter notebook notebooks/si_figures.ipynb
jupyter notebook notebooks/extra_figures.ipynb