CONCUR (version 1.2)

concur: Copy Number Profile Curve-Based Association Test

Description

Implements a kernel-based association test for CNV aggregate analysis in a certain genomic region (e.g., gene set, chromosome, or genome) that is robust to the within-locus and across-locus etiologoical heterogeneity, and bypass the need to define a "locus" unit for CNVs.

Usage

concur(cnv, X, pheno, phenoY, phenoType, ..., nCore = 1L,
  outFileKernel = NULL, verbose = TRUE)

Arguments

cnv

A character or data.frame object. If character, the name of the data file containing the CNV data (with a header). If data.frame, the CNV data. The data must contain the following columns: "ID", "CHR", "BP1", "BP2", "TYPE", where "ID" is a unique patient id, "CHR" is the CNV chromosome, "BP1" is the start location in base pairs or kilo-base pairs, "BP2" is the end location in base pairs or kilo-base pairs, and "TYPE" is the CNV copy number.

X

A character or data.frame object. If character, the name of the data file containing the covariate data (with a header). If data.frame, the covariate data. The data must contain a column titled "ID" containing a unique patient id. This column must contain the same patient identifiers as contained in the CNV data specified in input cnv. Categorical variables must be translated into design matrix format.

pheno

A character or data.frame object. If character, the name of the data file containing the phenotype data (with a header). If data.frame, the phenotype data. The data must contain a column titled "ID" containing a unique patient id. This column must contain the all of the patient identifiers contained in the CNV data specified in input cnv.

phenoY

A character object. The column name in input pheno containing the phenotype of interest.

phenoType

A character object. Must be one of of {"bin", "cont"} indicating if input phenoY (i.e., the phenotype of interest) is binary or continuous.

...

Ignored. Included to require named inputs.

nCore

An integer object. If nCore > 1, package parallel is used to calculate the kernel. Though the methods of package CompQuadForm dominate the time profile, setting nCore > 1L can improve computation times.

outFileKernel

A character object or NULL. If a character, the file in which the kernel is to be saved. If NULL, the kernel is returned by the function.

verbose

A logical object. If TRUE, progress information is printed to the screen.

Value

A list containing the kernel (or its file name) and the p-value.

Details

The CNV data must adhere to the following conditions:

• CNVs must be at least 1 unit long.

• CNVs cannot end at the exact location another begins

Violations of these conditions typically occur when data are rounded to a desired resolution. For example

 ID CHR      BP1      BP2 TYPE
  1  13 10112087 10112414    3

becomes upon rounding to kilo

 ID CHR   BP1   BP2 TYPE
  1  13 10112 10112    3     .

These cases should either be discarded or modified to be of length 1, e.g.,

 ID CHR   BP1   BP2 TYPE
  1  13 10112 10113    3     .

As an example of condition 2

 ID CHR    BP1   BP2 TYPE
  1  13 100768 101100    3
  1  13 101100 101299    1

should be modified to one of

 ID CHR    BP1   BP2 TYPE
  1  13 100768 101100    3
  1  13 101101 101299    1

or

 ID CHR    BP1   BP2 TYPE
  1  13 100768 101099    3
  1  13 101100 101299    1     .

Additionally,

 ID CHR    BP1   BP2 TYPE
  1  13 100768 101100    3
  1  13 101100 101299    3

should be combined as

 ID CHR    BP1   BP2 TYPE
  1  13 100768 101299    3     .

References

Brucker, A., Lu, W., Marceau West, R., Yu, Q-Y., Hsiao, C. K., Hsiao, T-H., Lin, C-H., Magnusson, P. K. E., Holloway, S. T., Sullivan, P. F., Szatkiewicz, J. P., Lu, T-P., and Tzeng, J-Y. Association testing using Copy Number Profile Curves (CONCUR) enhances power in copy number variant analysis. <doi:10.1101/666875>.

Examples

# NOT RUN {
data(cnvData)

# binary phenoType
results <- concur(cnv = cnvData,
                  X = covData,
                  pheno = phenoData,
                  phenoY = 'PHEB',
                  phenoType = 'bin',
                  nCore = 1L,
                  outFileKernel = NULL,
                  verbose = TRUE)

# continuous phenoType
results <- concur(cnv = cnvData,
                  X = covData,
                  pheno = phenoData,
                  phenoY = 'PHEC',
                  phenoType = 'cont',
                  nCore = 1L,
                  outFileKernel = NULL,
                  verbose = TRUE)

# }