In sean-cho/Epicopy: Get CNV information from 450K array

Running epicopy

The key function within Epicopy is the function epicopy which accepts the directory containing the raw Illumina .idat files and a sample sheet in a .csv format. The default argument of epicopy runs the code using all the normals included as part of the package (for more information, see Specifying normal samples below).

The function returns 1) circular binary segmentation results in R and 2) the same segmentation results and a marker file for running GISTIC2.0 as tab-delimited files in the current directory (unless otherwise specified using the output_dir argument). Note that for the purpose of space and formatting, the progress messages that Epicopy prints out will be suppressed in our examples.

library(Epicopy)
library(minfi)
data(data_vignette)
input_loc <- system.file('extdata', 'raw_idat', package = 'Epicopy')

epi_seg <- suppressMessages(epicopy(input_loc, output_dir = FALSE))

class(epi_seg)
head(epi_seg$output)

Starting from RGChannelSet

If the user desires to start from a pre-read RGChannelSet, the Epicopy package also includes individual functions that allow them to do so. This section will outline that process.

Load dummy data

First, we will read data from .idat files using the minfi package.

input_loc <- system.file('extdata', 'raw_idat', package = 'Epicopy')
epi_ss <- read.metharray.sheet(input_loc)
head(epi_ss)
epi_rg <- read.metharray.exp(targets = epi_ss)

epi_rg

Get log R ratios

Following that, we will run getLRR to obtain the log R ratios (LRR) of the samples compared to reference normals that were included in the Epicopy package. For this exercise we will be using the median values of the samples

epi_lrr <- getLRR(rgSet = epi_rg, Normals = NA)

head(epi_lrr)

Get CNA and segment data

A second function, LRRtoCNA, is used to generate segments from the LRR information from the previous step. This uses the DNAcopy package and ParDNAcopy package for parallelization. As such, LRRtoCNA includes arguments for input to pass to the CNA, smooth.CNA, and segment functions. See ?LRRtoCNA for more details.

epi_cna <- LRRtoCNA(epi_lrr)

class(epi_cna)
head(epi_cna$output)

This object is a CNA object which holds the segmented data in $output. At this point, the user can use this as any other segmented data for their analysis.

GISTIC 2.0

Epicopy also includes a wrapper function that allows the users to export both the segmented data and a marker file of the probes used in the segmentation process. Both files are the inputs for GISTIC 2.0 on the GenePattern server hosted by the Broad Institute.

There are three key arguments other than the segmented data; - output_dir: Output directory. Defaults to current. - filterbycount: Should a filter of minimum probes within a segment be included? - min_probes: If the previous is TRUE, what should the threshold be?

For the min_probes argument, we recommend 50 based on our experience with breast and lung data.

For our example, we will not evaluate the following chunk.

export_gistic(epi_seg, filterbycount = TRUE, min_probes = 50)

Visualize data

The plot_segments function is included in the package to allow the users to visualize the segmented data.

plot_segments(epi_seg, which_sample = 1)

Specifying normal samples

For epicopy function

User's input

If reference normal samples are included in the raw data files, a column specifying the normal status of the samples should be included. Normals have to be tagged using the character string normal (case insensitive).

EpicopyData/TCGA-derived normals

Otherwise, users can specify the one of the three type of normals included in the EpicopyData package, derived from normal solid tissue arrayed by the Cancer Genome Atlas (TCGA). To use those, users may input one of four arguments 'thyroid', 'breast', 'lung', or 'all', the last of which uses all available normals. The default uses all normal samples.

For getLRR function

Default

Defaults to NULL which uses all the normal samples included with the EpicopyData package.

EpicopyData/TCGA-derived normals

To use EpicopyData included normals, as before, normals can be specified using one of four arguments.

User's input

If the user has their own normal samples, they can specify either a numeric/integer index that identifies the positions of the normal samples in the RGChannelSet or a logical vector that flags normal samples as TRUE.

Use all samples

The argument Normals = NA uses the mode/median (as specified by the user) of all the samples, regardless of status, as reference normals. The idea behind this is that the median copy number of a given genomic region of all the samples should center around zero. Recommended only when there are many samples in the array.

Session information

sessionInfo()

sean-cho/Epicopy documentation built on May 29, 2019, 4:24 p.m.