Plot parameters of the variable length Markov model (VLMM) for read starts

Description

This function plots portions of the Cufflinks VLMM for read start bias. The natural log of observed over expected is shown, such that 0 indicates no contribution of a position to the read start bias. As the variable lenght Markov model has different dependencies for different positions (see Roberts et al, 2011), it is difficult to show all the 744 parameters simultaneously. Instead this function offers to show the 0-order terms for all positions, or the 1st and 2nd order terms for selected positions within the read start sequence. For the 1- and 2-order terms, the log bias is shown for each nucleotide (A,C,T,G) given the previous nucleotide (1-order) or di-nucleotide (2-order).

Usage

1
2
3
4
5
plotOrder0(order0, ...)

plotOrder1(order1, pos1)

plotOrder2(order2, pos2)

Arguments

order0

the "order0" element of the list named "vlmm.fivep" or "vlmm.threep" within the list that is the output of fitBiasModels

...

parameters passed to plot

order1

as for "order0" but "order1"

pos1

the position of the 1st order VLMM to plot

order2

as for "order0" but "order2"

pos2

the position of the 2nd order VLMM to plot

Value

plot

Functions

  • plotOrder1: Plot first order parameters for a position

  • plotOrder2: Plot second order parameters for a position

References

Roberts et al, "Improving RNA-Seq expression estimates by correcting for fragment bias" Genome Biology (2011) doi:101186/gb-2011-12-3-r22

Examples

1
2
3
4
5
6
7
# fitpar was fit using identical code
# as found in the vignette, except with
# 25 genes, and with fragment size in 80-350 bp
data(preprocessedData)
plotOrder0(fitpar[[1]][["vlmm.fivep"]][["order0"]])
plotOrder1(fitpar[[1]][["vlmm.fivep"]][["order1"]], pos1=5:19)
plotOrder2(fitpar[[1]][["vlmm.fivep"]][["order2"]], pos2=8:17)