rhngla/cplAE_snmCAT

Coupled autoencoders snmCAT dataset

Environment

1. Create the conda environment with the required dependencies.

conda create -n snmCAT
conda activate snmCAT
conda install python=3.7
conda install pytorch torchvision torchaudio -c pytorch
pip install jupyterlab scikit-learn feather-format seaborn pandas rich tqdm timebudget autopep8 pyqt5

2. Clone and install this repo:

git clone https://github.com/rhngla/cplAE_snmCAT
cd cplAE_snmCAT
pip install -e .

3. Get the data required for this repository

data_dir
    ├── CH.csv
    ├── mCH.csv
    ├── metadata.csv
    └── rna.csv

4. Specify data paths in config.toml (at repository root level):

data_dir = "/home/Local/dat/raw/snmCAT-seq/"
metadata_file = "/home/Local/dat/raw/snmCAT-seq/metadata.csv"
rna_file = "/home/Local/dat/raw/snmCAT-seq/rna.csv"
mCH_file = "/home/Local/dat/raw/snmCAT-seq/mCH.csv"
CH_file = "/home/Local/dat/raw/snmCAT-seq/CH.csv"
outlier_file = "/home/Local/dat/raw/snmCAT-seq/outliers.csv" # .csv is generated within notebook 02_ouliers_E.ipynb

Dataset description

Human snmCAT-seq dataset from Luo et al. 2019, shared by Fangming Xie from Eran Mukamel's group.

DNA methylation (mC), open chromatin (A), and transcriptomes (T) were measured in the same set of single nuclei .

• metadata.csv: general information of each cell, including ID, biological sample, and cell cluster labels.
• rna.csv: Unnormalized cell x gene count matrix
• mCH.csv: cell x gene matrix, representing methylated cytosine count in the gene body of that gene
• CH.csv: cell x gene matrix with total number of cytosines (methylated + unmethylated).

mCH and CH matrices have the same dimensions. Element-wise ratio of mCH to CH (i.e. fraction of methylated to total cytosines) are usually interpreted as DNA methylation levels.