CellFindPy is a program that uses Scanpy (https://github.com/theislab/scanpy) to cluster and further subcluster cells by single cell RNA sequencing 10x Data in an automated fashion.
CellFindPy analysis 10x Data in the following steps.
-
First it initializes an Anndata (https://github.com/theislab/anndata) object and filters out cells and genes as described in de Scanpy api. For example, you can set to filter out cells that have x% their gene expression assigned to mitochondrial genes.
-
Secondly, it calculates pca, neighbor graph and find clusters using Scanpy (louvain clustering).
-
After clustering, marker genes are identified using both average differentiated expression and Mann–Whitney U test (to calculate p-values).
-
With this information a cluster test is conducted to identify if the set resolution gives distinct enough clusters. The thresholds for the cluster test is fully custumizable. As an example, here we used a threshold of at least 10 genes that are >log(2) differentially expressed between clusters.
-
Lastly, looping through this system is done to find the highest possible resolution where the thresholds are met. Further subclustering and even subsubclustering can be done using this same 5 step procedure
Importantly, filters and threshold are fully custumizable in our scripts and can be interactively setup at the start of initializing using the setup.sh.
Alternatively, clustering can be initialized using the run_script.py and seperate functions can also be used by import Lib_.py.
To be determined
First step of install is only required to be able to use both setup.sh and run_script.py. If you want to use functions separately than it is advised to use pip.
git clone https://gitlab.com/Douwe-Spaanderman/CellFindPy
cd CellFindPy/
pip install .
Initializing using setup.sh should install all required packages used in CellFindPy.
To run open terminal:
cd /CellFindPy/CellFindPy #locate _setup_.sh
bash _setup_.sh
Follow the instructions given by setup.sh to initialize CellFindPy. Alternatively:
cd /CellFindPy/CellFindPy #locate run_script.py
python3 run_script.py -h
-h will show you all the possible arguments to set. Only the input folder (with the matrix.mtx, genes.tsv and barcodes.tsv), output folder and save name are required.
To be determined
To be determined