combineCGandGCmatrices: Combine CG and GC methylation matrices
In jsemple19/methMatrix: Manipulate Lists of Methylation Matrices

combineCGandGCmatrices

R Documentation

Combine CG and GC methylation matrices

Description

Methylation matrices are tricky to merge because:

Usage

combineCGandGCmatrices(matCG, matGC, regionGR, genomeMotifGR)

Arguments

`matCG`	Methylation matrix (reads x positons) of C positions within CG motifs
`matGC`	Methylation matrix (reads x positons) of C positions within GC motifs
`regionGR`	GRanges object of region used to make methylation matrices
`genomeMotifGR`	GRanges object with all unique non-overlapping CG/GC/GCGorCGC motifs in genome

Details

1. Forward and reverse reads will have methylation calls at different positions even if they are part of the same motif because the C is shifted by 1 between the two strands. Therefore we "destrand" the reads by averaging all methylation calls within the same motif (and ingoring NAs). The genomic positions in the final merged matrix will be on the position of the 1st bp of the motif for CG motifs and the genomic position of the 2nd bp of the motif for GC motifs.

2. GCG/CGC motifs will have methylation calls from three Cs and it is not clear whether the middle C should be part of one motif or the other. Therefore we consider such triplets as a single motif and average the methylation calls from all three Cs. The genomic position used in the final merged matrix will be that of the middle bp of the motif.

3. Longer GCGC/CGCG runs have multiple overlapping motifs layered within them. To deal with that we split these runs into 2-3bp motifs. If the length of the run is an even number, it is split into doublets and considered as simple CG or GC motifs. If the length of the run is an odd number, one triplet is created, and the rest is split into doublets. This creates a set of unique non-overlapping motifs in the genome that are either CG or GC 2bp motifs, or a GCG/CGC 2bp motif. This unique set is created before hand with the nanodsmf::findGenomeMotifs function and can be stored in a RDS file. Doublet and triplet motifs created from the run are treated the same as isolated doublet and triplet motifs, as described in points 1. and 2. above.