Get gene body coverage

geneBodyCoverage(obj, ...): calculate gene body coverage from input

Calculate geneBodyCoverage of query sequences

Usage

geneBodyCoverage(obj, ...)

# S4 method for AlignmentPairs
geneBodyCoverage(obj, min.match = 0.9)

# S4 method for AlignmentPairsList
geneBodyCoverage(obj, min.match = 0.9, bpparam = NULL)

Arguments

obj

AlignmentPairs object

...

additional parameters

min.match

filter out hits with fraction matching bases less than min.match

bpparam

BiocParallel parameter object

Value

reduced and filtered DataFrame object where each row is a transcript. See @details for information on seqnames and seqlengths columns are collected from the corresponding seqinfo object. breadthOfCoverage is the total width of all reduced regions of the transcript and corresponds to how much of a transcript is covered. Dividing breadthOfCoverage by the total transcript length gives the coverage fraction. depthOfCoverage sums all ranges and is an estimate of how many times each range is present. Dividing the depthOfCoverage by breadthOfCoverage indicates the multiplicity of the query. For queries that map multiple subjects, a value close to 1 indicates the query has been split in several subjects, whereas a higher value indicates sequence duplication at the subject level.

The revmap column maps the output ranges to the input ranges
as lists of numerical ids. These ids can be used to retrieve
the corresponding AlignmentPairs ranges providing a link to
the subjects. The hitCoverage lists the width of each reduced
hit, and hitStart and hitEnd provide the transcript
coordinates of these hits.

Details

Given an AlignmentPairs object, calculate gene body coverages of query sequences. regions. For instance, if one hit spans coordinates 100-400 and another 200-500, we merge to 100-500. The function calculates how many overlaps that are merged, the rationale being that if there are >1 overlaps for a seqname, the sequence is associated with multiple distinct regions in the subject sequence. Thus, the unit of interest here is the query, and the function seeks to identify reduced disjoint regions of a query sequence that map to a subject.

The function returns a GRanges object with reduced regions,
i.e. overlaps are merged. Information about matches and
mismatches is currently dropped as it usually is not possible
to infer where mismatches occur. Instead, the data column
`coverage` simply holds the ratio of the width of the region
to the transcript length. Summing up coverages from disjoint
regions then gives total coverage of the transcript.

The cutoff is used to filter regions based on the ratio of
matches to the width of the region.

Since the association between query and subject regions is
removed, the return value is a GRanges object consisting of
the reduced query ranges with a revmap and coverage attribute.