coverageScorings: Add a coverage scoring scheme

In Roleren/ORFik: Open Reading Frames in Genomics

Add a coverage scoring scheme

Description

Different scorings and groupings of a coverage representation.

Usage

coverageScorings(coverage, scoring = "zscore", copy.dt = TRUE)

Arguments

coverage	a data.table containing at least columns (count, position), it is possible to have additionals: (genes, fraction, feature)
scoring	a character, one of (zscore, transcriptNormalized, mean, median, sum, log2sum, log10sum, sumLength, meanPos and frameSum, periodic, NULL). More info in details
copy.dt	logical TRUE, copy object, to avoid overwriting original object. Set to false to run function using reference to object, a speed up if original object is not needed.

Details

Usually output of metaWindow or scaledWindowPositions is input in this function.

Content of coverage data.table: It must contain the count and position columns.

genes column: If you have multiple windows, the genes column must define which gene/transcript grouping the different counts belong to. If there is only a meta window or only 1 gene/transcript, then this column is not needed.

fraction column: If you have coverage of i.e RNA-seq and Ribo-seq, or TCP -seq of large and small subunite, divide into fractions. Like factor(RNA, RFP)

feature column: If gene group is subdivided into parts, like gene is transcripts, and feature column can be c(leader, cds, trailer) etc.

Given a data.table coverage of counts, add a scoring scheme. per: the grouping given, if genes is defined, group by per gene in default scoring.
Scorings:

• zscore (count-mean(window))/sd(window) per) # Outlier detection

• modzscore (count-median(window))/mad(window) per) # Outlier detection for signal with extreme values

• transcriptNormalized (sum(count / sum of counts per)) this is scaled per group

• mean (mean(count per))

• median (median(count per))

• sum (count per)

• log2sum (count per)

• log10sum (count per)

• sumLength (count per) / number of windows

• meanPos (mean per position per gene) used in scaledWindowPositions

• sumPos (sum per position per gene) used in scaledWindowPositions

• frameSum (sum per frame per gene) used in ORFScore

• frameSumPerL (sum per frame per read length)

• frameSumPerLG (sum per frame per read length per gene)

• fracPos (fraction of counts per position per gene) non scaled version of transcriptNormalized

• periodic (Fourier transform periodicity of meta coverage per fraction)

• NULL (no grouping, return input directly)

Value

a data.table with new scores (size dependent on score used)

See Also

Other coverage: metaWindow(), regionPerReadLength(), scaledWindowPositions(), windowPerReadLength()

Examples

dt <- data.table::data.table(count = c(4, 1, 1, 4, 2, 3),
                             position = c(1, 2, 3, 4, 5, 6))
coverageScorings(dt, scoring = "zscore")

# with grouping gene
dt$genes <- c(rep("tx1", 3), rep("tx2", 3))
coverageScorings(dt, scoring = "zscore")

Roleren/ORFik documentation built on Dec. 18, 2024, 11:39 p.m.