ENCODE Single-Cell Working Group: Analyses for single-nucleus ATAC-seq data

snATAC-seq processing and manual annotation pipeline for part of ENCODE single-cell data collection Please see the Experiment-ID.txt for more information.

Please contact Yu Fu (yu.fu2@umassmed.edu) if you have any question!

Environment setup

• install conda
• install python
• conda env create -f environment.yml

Steps

Step 0: Download the fragment file from ENCODE with authetication

python downloadData.py $EXP $File $User $pswd
tar -xvf $EXP.tar.gz

Fragment File path: encode_scatac_dcc_2/results/[EXP]-1/fragments/fragments.tsv.gz )

Step 1: Build ArchR Project

Rscript run-ArchR.R $file $EXP $biosampleID
mkdir $EXP/MarkerFeatures

Step 2: Filtering out subset of all-tissue marker gene list

Can add "PanglaoDB" for more canonical marker genes)

awk -v tissue=$tissue '{if (($2==tissue) || ($2=="multiple")) print $1}' > markers

two ways of cell type annotation

Step 3A: Directly overlap marker genes list with marker features passing a certain FDR and Log2FC threshold

Rscript Pull-Marker-Gene.R $EXP/Save-ArchR-Project.rds $EXP/MarkerFeatures
./predict-cell-type-normax.sh $EXP/MarkerFeatures/ markers > overlap-results

Step 3B: Visulize the raw or transformed gene score for annotation

Rscript Draw-UMAP.R $EXP/Save-ArchR-Project.rds $EXP
Rscript Draw-Marker-Heatmap.R $EXP/Save-ArchR-Project.rds markers $EXP
Rscript Draw-Dot-Plot $Exp/Save-ArchR-Project.rds markers $EXP

Step 4: Assign labels

Saved in proj$Labels and write to master table

Rscript Draw-UMAP-with-Label.R $EXP/Save-ArchR-Project.rds $EXP