Introduction

GIML produces segmentations of methylation data generated by whole genome bisulfite sequencing (WGBS). The coarsness of the segmentation is controlled by a parameter called $\lambda$. The algorithm behind GIML is similar to the one used in Vega (https://doi.org/10.1093/bioinformatics/btq586) but it uses a different probabilistic model tailored for WGBS.

Usage

giml <input-file> <lambdas>

input file is a file in tab-separated format:

chromosome position non-converted-reads converted-reads if the file name is '-' giml will read from the standard input.

One can specifiy more tha one lambda by using ":" as a separator. The different lambdas will be tried in sequence.

Output format

tab-separated format:

chromosome segment-start segment-end lambda index-start index-end methylation-mle

where index-start and index-end are indexes for the segment into a zero-based array which stores the positions of the chromosome (useful for plotting and counting the CpGs contained in a given segment).

Other scripts bundled with giml

the giml repository contains scripts to perform basic descriptive statistics on giml output, and to help plotting the segments.