germlineFilters: Identify focal, somatic hemizygous deletions and somatic...

In cancer-genomics/trellis: Somatic structural variant analysis

Identify focal, somatic hemizygous deletions and somatic homozygous deletions

Description

Segmentation of bin-level normalized log2 ratios yields a set of genomic intervals each having an associated mean. We define candidate somatic deletions as those having a segment mean less than a specified cutoff (e.g., the theoretical segment mean of a hemizygous deletion on log2 scale). For each candidate CNV, we assess whether the variant can be explained by events identified in germline samples processed in the same batch, or by sequence artifacts (low mappability and/or GC). To help ensure that the deletion is focal (not a small deletion embedded in a larger deletion), we (i) verify that the candidate CNV (cnv) has a mean log ratio less than the mean log ratio of the neighboring segments by an amount of approx. -0.5 [log2(0.7)] by default and (ii) assess whether the interval could plausibly be a large hemizygous deletion. Candidate somatic deletions identified by this function have the following properties: (i) not germline, (ii) unlikely to be large hemizygous deletions, (iii) the difference in mean of the segment and neighboring segments is less than -0.5 (by default), and (iv) have a width of at least 2kb (by default).

Usage

germlineFilters(preprocess, germline_filters, param = DeletionParam())

Arguments

preprocess	a list of preprocessed data. See preprocessData
germline_filters	A GRanges object (e.g., germline CNVs and regions of low sequence quality)
param	A DeletionParam object

Value

a named GRanges object. The names are given by paste0("sv", seq_along(g)) where g is the GRanges object.

See Also

preprocessData

cancer-genomics/trellis documentation built on Aug. 20, 2024, 5:48 p.m.