Uniwig

Given a set of bed files, we want to produce 3 BigWig output files: one track of the start coordinates, one track of the end coordinates, and one track for core coordinates.

Prerequisites

Uniwig will rely on libBigWig, a C library for processing BigWig files. You will need to compile the libBigWig library locally first, then you can compile uniwig. The steps are described below.

Compiling uniwig

1. Clone the libBigWig and uniwig repositories

We recommend you clone this repository so libBigWig and uniwig are parallel folders at the same level.

git clone https://github.com/dpryan79/libBigWig
git clone https://github.com/databio/uniwig

2. Compile libBigWig with provided Makefile

This should add two library files libBigWig.a and libBigWig.so in your libBigWig local repository:

cd libBigWig
make install prefix=lib

3. Compile uniwig

With libBigWig compiled, you can use the Makefile to compile, test, or remove uniwig. The compiled program will have the path of ./bin/uniwig (assuming you are already in the uniwig local repository). Below are some commands with different Makefile usage.

To compile uniwig:

cd uniwig
make uniwig

To test uniwig:

make tests

To remove uniwig (remove ./bin/uniwig):

make clean

To recompile a new version of uniwig after change:

make rebuild

If your libBigWig and uniwig local repositories are not located at the same level, then change the LIB_DIR in the Makefile. The default path for LIB_DIR is a relative path from uniwig local repository to libBigWig local repository, which has the library files already compiled.

Uniwig usage:

To use the compiled uniwig program located at ./bin/uniwig, use the command with the following format

./bin/uniwig (-s) -m (5) -w (1) $combined_bed_file_path $chrom_size_file_path $bw_output_file_header

With these parameters:

• -s for specifying whether the provided combined bed file is already sorted by the chromosome number. If -s flag is not given, uniwig will sort the combined bed file by chromosome number.
• -m for specifying the smooth size. The positive integer number given after the -m flag will be used to select the window size for smoothing the coordinates.
• -w for specifying the write size. The data will be written into .bw files in chunks to reduce memory usage, and the chunk size (number of lines) will be determined by the -w flag.
• $combined_bed_file_path for specifying the path of combined bed file. A relative path given would start changing directories from uniwig local repository (it is recommended to put the combined bed file in ./data/combined/)
• $chrom_size_file_path for specifying the path of chromosome size reference file. There is a reference file provided in this repository, located at ./test/hg38.chrom.sizes, but you are welcome to use any other reference files (such as refgenie)
• $bw_output_file_header for specifying the header of the BigWig output file. There will be three .bw files produced, ending with _start.bw, _end.bw, and _core.bw. The input for this parameter will be added to the output file paths as prefixes (it is recommended to put the output BigWig file in ./data/bw/)

You can also use the provided shell script for executing the entire workflow: from a set of raw data (bed files) to the final BigWig file output. There are two choices provided in this repository:

• create_unsorted.sh is the shell script for executing uniwig without sorting the combined bed file by chromosome number before it's submitted to uniwig. In other words, uniwig will be responsible for sorting the chromosomes.
• create_sorted.sh is the shell script for executing uniwig with sorting the combined bed file by chromosome number before it's submitted to uniwig. The sorting is done by shell script command sort.

For both shell script, the default paths for the required files are provided. It is recommended to organize your input files by the default choice, but make sure to change the paths to corresponding repositories if you are not using the default paths.