Installation

cvlr depends on htslib. You need to install it and write the relevant path in the Makefile. Specifically look at the lines starting with INCLUDE and LIB. If you have installed htslib in the directory /home/user/src/htslib those two lines will have to be:

INCLUDE=-I/home/user/src/htslib/htslib
LIB=/home/user/src/htslib/libhts.a -lpthread -lz -lcurl -llzma -lbz2 -lcrypto -lm

The htslib has to be recent enough as to contain functions to parse the Mm/Ml tags. I don't know what the oldest useful version is, but I'm sure 1.16 works.

Once you have edited the Makefile, just type make.

Usage

You need to have a BAM/CRAM file containing methylation data encoded with the Mm/Ml tags as explained in the SAM documentation (https://samtools.github.io/hts-specs/SAMtags.pdf). This is also the standard output format (at the moment) of Nanopore's Megalodon. First you create the matrix from the BAM file with

cvlr-meth-of-bam NA12878.cram chr20:58850594-58852978 > GNAS-matrix.txt

then you cluster the reads with

cvlr-cluster GNAS-matrix.txt 2 1 100 > GNAS-clusters.txt

(That is you are creating 2 clusters, with seed 1 for the random numbers generator and a maximum of 100 EM iterations).

You can then look at the clusters (for example for plotting the methylation values) with

cvlr-stats.py GNAS-clusters.txt GNAS-matrix.txt > GNAS-stats.txt

If you have 2 clusters columns 1,4,7 of contain the genomic position, methylation in cluster 0 and methylation in cluster 1 respectively.