vafClust: punishingly simple VAF clustering for clonality estimation

In ttriche/sesamizeGEO: the poor man's methylation suite (cf. `sesamizer`)

punishingly simple VAF clustering for clonality estimation

Description

Suppose we have read coverage over a base with a detected variant. We can model the reads mapping to variant and reference bases as

Usage

vafClust(alt, ref, depth = NULL, ...)

Arguments

alt	read depth for the variant allele
ref	read depth for the reference allele
...	additional arguments to pass on to dbscan
eps	epsilon to use for DBSCAN clustering

Details

p ~ Beta(variant_reads, reference_reads)

Now, if we take logits, with nu = logit(p), then

nu ~ Normal( log(p/(1-p)), log10(total reads) )

As a bonus, we can model nu as roughly multivariate normal. Furthermore, we can use the latter term as observation weights. Given a vector or matrix of VAFs, this function clusters them using dbscan, with log10(read_depth) as the weight for each VAF.

Value

   an object of class `dbscan_fast`

ttriche/sesamizeGEO documentation built on Nov. 12, 2023, 5:42 p.m.