Note : The lastest version of screw can be found at https://github.com/Epigenomics-Screw/Screw
A CWL Pipeline and Docker Image for Performing Standard Single-Cell DNA Methylation Analyses
Single-cell whole-genome bisulfite sequencing (SC-WGBS) is currently the most comprehensive method for investigating single-cell epigenetic regulation, with important applications for understanding stem cell differentiation and oncogenesis. However, it is a very new field, and while a handful of tools have been developed, no standard pipeline or framework for handling this data has been created, despite a number of analyses being common across the studies which have been published so far.
The purpose of this project is to provide a standardised workflow for analysing single-cell DNA methylation data. We do this by wrapping several existing tools to create Common Workflow Language modules. All modules are provided as part of an executable-anywhere Docker image. We include a small subset of toy data from single cells published by Farlik et al. 2015 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE65196). The planned, complete pipeline is viewable below and as units of the workflow are implemented and tested as CWL tools, they will be colored in.
You can use whichever CWL runner you prefer. We tested on CWLref-runner and cwl tool. To install them, you can use pip. If you do not have pip, you can check out cwltool directly from their github: https://github.com/common-workflow-language/cwltool. NOTE All these tools are python-2.7 branch ONLY. These tools WILL NOT work with python3+.
pip install cwlref-runner
pip install cwltool
With Docker, you don't need to install anything else. Without Docker, you need to install all of the programs which are listed in Docker/Dockerfile.
After all input files are formatted properly, invoke the master CWL script below:
cwltool SCREW.cwl
This will currently run through the workflow we implemented in the NCBI hackathon: 
Users should place raw single-cell methylation count data of the type stored in GEO accession GSE65196 in a working directory and be prepared for the size of the data to as much as triple during the course of the run.