Installation and configuration¶
conda install segway
pip install segway
If you are using SGE, your system administrator must set up a
mem_requestedresource for Segway to work. This can be done by installing Segway and then running
python -m segway.cluster.sge_setup.
Observation data is stored with the genomedata system. <http://pmgenomics.ca/hoffmanlab/proj/genomedata/>. There is a small Genomedata archive for testing that comes with Segway, that is used in the below steps. You can get it using:
segway traincommand to discover patterns in the test data. Here, we specify that we want Segway to discover four unique patterns:
segway --num-labels=4 train test.genomedata traindir
segway identifycommand to create the segmentation, which partitions the genome into regions labeled with one of the four discovered patterns:
segway identify test.genomedata traindir identifydir
This example spawns jobs that will run sequentially due to small
segment size. See the
--split-sequences option for
dividing segments into smaller pieces.
identifydir/segway.bed.gzfile has each segment as a separate line in the BED file, and can be used for further processing.
identifydir/segway.layered.bed.gzfile is designed for easier visualization on a genome browser. It has thick lines where a segment is present and thin lines where it is not. This is not as easy for a computer to parse, but it is more useful visually.
You can also perform further analysis of the segmentation and trained parameters using Segtools <http://pmgenomics.ca/hoffmanlab/proj/segtools/>.