The goal of this repository is to give some simple shell scripts to process WGBS data, ultimately to be used for downstream analysis (especially with CelFiE)
To run this code the following software is needed:
- Python >= 3.0
- Bedtools
- The
bigWigToBedGraphcommand line utility from UCSC
The scripts run on the command line like this:
./process_encode.sh
./process_blueprint.sh
The output should be one file, with a total of number of ~28 million sites.
To convert to hg38, take a final bed file (with no header) and use the liftover python script in this repository
python liftover.py <input file name> <output file name> <delimiter of file>
For example, if you had a tab delimited file named "adipose.bed", you would run
python liftover.py adipose.bed adipose_hg38.bed tab
This code only works for hg19->hg38. To change it, you need to change the UCSC liftover chain file specified in the python script.
There will likely be slightly difference in the number of final sites versus the starting number, but if this difference is larger than 10% of the starting number, there was likely something wrong with the conversion.
If there is more than one replicate of a tissue, and you wish to combine them, use the following python script:
python combine.py <rep1 bed file> <rep2 bed file> <output file>
ex:
python combine.py brain1.all_sites brain2.all_sites brain_all.meth_depth
See process_encode.sh for an example in a bash context.
To subset to preexisting TIMs (calculated using a set of ~20 tissues taken from ENCODE and BLUEPRINT), follow tims-scripts/intersect_tims.sh