splicepca: SplicePCA: splicing graph PCA plot

In pkimes/spliceclust: Visualization and Clustering of Spliced Components

Description

Splice graphs with PCA loadings shown along the exons and junctions to illustrate expression patterns across a connected component. The function can be used to illustrate, e.g. differences across groups/clusters or principal component loadings. Note that PCA is performed on the log-transfored data by default.

Usage

splicepca(obj, npc = 3, pc_sep = TRUE, ej_w = c(1, 1), log_base = 10,
  log_shift = 1, genomic = TRUE, ex_use = 2/3, flip_neg = TRUE,
  use_blk = FALSE, txlist = NULL, txdb = NULL, orgdb = NULL,
  scores = FALSE, plot = TRUE, ...)

## S4 method for signature 'concomp'
splicepca(obj, npc = 3, pc_sep = TRUE, ej_w = c(1, 1),
  log_base = 10, log_shift = 1, genomic = TRUE, ex_use = 2/3,
  flip_neg = TRUE, use_blk = FALSE, txlist = NULL, txdb = NULL,
  orgdb = NULL, scores = FALSE, plot = TRUE, ...)

Arguments

obj	a concomp object with exon and junction information
npc	a numeric value specifying number of PCs to plot (default = 3)
pc_sep	a logical whether PCA should be performed on exon and junction coverage separately (default = TRUE)
ej_w	a numeric vector of length two specifying the relative sum of squares for exon and junctions (default = c(1, 1))
log_base	a numeric specifying the scale of the expression values at each exon, which 0 resulting in no log scaling being applied (default = 10)
log_shift	a numeric specifying the shift to be used in the log transformation for handling 0 values (default = 1)
genomic	a logical whether genomic coordinates should be used to plot the heatmap (default = TRUE)
ex_use	a numeric specifying the proportion of the plot exons should occupy if non-genomic coordinate plotting is desired (default = 2/3)
flip_neg	a logical whether to flip plotting of genes on negative strand to run left to right (default = TRUE)
use_blk	a logical whether to use a black background (default = FALSE)
txlist	a GRangesList of transcripts or genes which should be queried and added to the plot if falling within the region of the connected component (default = NULL)
txdb	a transcript database which can be used to query the transcript IDs identified from txlist (default = NULL)
orgdb	a database that can be queried using keys obtained from txdb to determine corresponding gene symbols (default = NULL)
scores	a logical whether to produce score plot rather than loadings plot (default = FALSE)
plot	a logical whether to output a plot or the actually PCA analyses performed by prcomp (default = TRUE)
...	other parameters to be passed

Details

npc cannot be larger than the rank of the centered data. When pc_sep = TRUE, this corresponds to min(p_e, p_j, n-1), where p_e, p_j, n correspond to the number of exons, junctions, and samples. When pc_sep = TRUE, this corresponds to min(p_e+p_j, n-1).

Value

a ggplot2 plot showing the specified number of loadings

Author(s)

Patrick Kimes

pkimes/spliceclust documentation built on Jan. 2, 2020, 4:44 a.m.