mpsmooth {mpss}R Documentation

Multi-Platforms Segmentation for Detection of Copy Number Variations

Description

Multi-Platforms segmentation for detection of CNVs

Usage

mpsmooth(pos1 = pos1, pos2 = pos2, pos3 = NA, pos4 = NA, data1 = data1,
data2 = data2, data3 = NA, data4 = NA, maxiter = 20, lambda = NA, lambda.range = c(20,600),
del.lim = -0.05, dup.lim=0.05, min.probes = 10, min.probe.density = 5,
min.length = 1000, min.dist = 5000, fdr.limit = 1e-05)

Arguments

pos1 Probe positions for platform 1. A list where each component corresponds to each chromosome. See Details.
pos2 Same as pos1, but for platform 2.
pos3 Same as pos1, but for platform 3. Set to NA if platform 3 is unavailable
pos4 Same as pos1, but for platform 4. Set to NA if platform 4 is unavailable.
data1 Normalized intensity ratio for platform 1, corresponding to each position in pos1. A list where each component corresponds to each chromosome. See Details.
data2 Same as data1, but for platform 3.
data3 Same as data1, but for platform 4. Set this to NA if there is no platform 3 available.
data4 Same as data1, but for platform 3. Set this to NA if there is no platform 4 available.
maxiter Maximum number of iterations in the iterative backfitting algorithm.
lambda A user defined value that determines the degree of smoothness, high value of lambda means we impose a smoother estimate. Set this to NA if one wishes to use AIC criteria to choose the optimal lambda
lambda.range The range of lambda to search for optimal lambda using AIC criteria.
del.lim Threshold for deletions. See Details.
dup.lim Threshold for duplications. See Details.
min.probes Minimum number of probes in region.
min.probe.density Minimum number of probes per kb in region.
min.length Minimum length of CNV regions (in bases).
min.dist Two regions will be merged if the distance between them is less than min.dist (in bases).
fdr.limit Fdr threshold to define significant segments. See Details.

Details

This is a convenient wrapper function that performs smoothing using smoothseg2 for all the chromosomes present in the input and then performs segmentation using segmentation, and returns regions that pass the user specified thresholds.\ Users are advised to first perform background normalization to minimize any platform-specific effects. For data from germline DNA, we recommend that users apply srsmooth to the unnormalized data in a chromosome-by-chromosome fashion, and use the 'residual' as input to this function.\ Deletions are sets of consecutive probes for which the smoothed intensities ars consistently smaller than or equal to del.lim. Duplications are sets of consecutive probes for which the smoothed intensities ars consistently greater than or equal to dup.lim. False Discovery rates are calculated from the p-values using Benjamini and Yekutieli's method. See p.adjust.

Value

cnv A data frame with the following columns: start: Start position of CNV regions. end: End position of CNV regions. chr: Chromosome number of CNV regions. start.loc: Index of start position in unlist(pos). end.loc: Index of end position in unlist(pos). length: length of region. p: p-value. fdr: false discovery rate. numprobes: number of probes in the region. scaled_x2: scaled chi squared value. cn: copy number status, 1 for deletion and 3 for duplication. m: mean of the probe intensities in the region. sd: standard deviation of the probe intensities in the region. pt: p value for test of discrepant regions. t_fdr: FDR for test of discrepant regions. m1: mean of the probe intensities from platform 1 in the region. m2: same as m1 but for platform 2. m3: same as m1 but for platform 3. m4: same as m1 but for platform 4. sd1: standard deviation of the probe intensities from platform 1 in the region. sd2: same as sd1 but for platform 2. sd3: same as sd1 but for platform 3. sd4: same as sd1 but for platform 4.
y Smooth segmented values corresponding to each position in pos.
pos Union of the probe positions in the platforms used.
chi probe-specific contribution to chi-squared statistic.

See Also

smoothseg2,segmentation

Examples

setwd(paste(searchpaths()[grep("mpss",searchpaths())],'/data/',sep=""))
load("illum12056.Rdata") #Illumina platform
#norm_y contains the normalized intensity ratios for chromosome 1 and 2.
#norm_x contains the corresponding probe locations.
illum = norm_y
illumx = norm_x
load("affy12056.Rdata") #Affymetrix platform
affy = norm_y
affyx = norm_x
s = mpsmooth(pos1 = affyx, pos2 = illumx, pos3 = NA, pos4 = NA, data1 = affy,
data2 = illum, data3 = NA, data4 = NA, maxiter = 50, lambda = 100, lambda.range = c(20,600)
, del.lim = -0.05, dup.lim=0.05, min.probes = 10, min.probe.density = 5,
min.length = 1000, min.dist = 5000, fdr.limit = 1e-05)  

[Package mpss version 1.2 Index]