Function to obtain SVD-based gene significance from the input gene-sample matrix

Description

dSVDsignif is supposed to obtain gene signficance from the given gene-sample matrix according to singular value decomposition (SVD)-based method. The method includes: 1) singular value decomposition of the input matrix; 2) determination of the eigens in consideration (if not given); 3) construction of the gene-specific project vector based on the considered eigens; 4) calculation of the distance statistic from the projection vector to zero point vector; and 5) based on distance statistic to obtain the gene significance.

Usage

dSVDsignif(data, num.eigen = NULL, pval.eigen = 0.01, signif = c("fdr", "pval"), 
  orient.permutation = c("row", "column", "both"), num.permutation = 100, fdr.procedure = c("stepup", 
          "stepdown"), verbose = T)

Arguments

data
an input gene-sample data matrix used for singular value decomposition
num.eigen
an integer specifying the number of eigens in consideration. If NULL, this number will be automatically decided on based on the observed relative eigenexpression against randomised relative eigenexpression calculated from a list (here 100) of permutated input matrix
pval.eigen
p-value used to call those eigens as dominant. This parameter is used only when parameter 'num.eigen' is NULL. Here, p-value is calcualted to assess how likely the observed relative eigenexpression are more than the maximum relative eigenexpression calculated from permutated matrix
signif
the singificance to return. It can be either "pval" for using the p-value as the gene significance, or "fdr" for using the fdr as the gene significance
orient.permutation
the orientation of matrix being permutated. It can be either "row" to permutate values within each row, or "column" to permutate values within each column, or "both" to permutate values both within rows and columns. Notably, when using the p-value as the gene significance, it is always to permutate values within each row.
num.permutation
an integer specifying how many permutations are used
fdr.procedure
the procedure to adjust the fdr. To ensure that the high distance statistic the more significance, the fdr should be adjusted either using "stepup" for step-up procedure (from the most significant to the least significant) or using "stepdown" for step-down procedure (from the least significant to the most significant)
verbose
logical to indicate whether the messages will be displayed in the screen. By default, it sets to true for display

Value

a vector storing gene significance

Note

none

Examples

# 1) generate data with an iid matrix of 1000 x 9 data <- cbind(matrix(rnorm(1000*3,mean=0,sd=1), nrow=1000, ncol=3), matrix(rnorm(1000*3,mean=0.5,sd=1), nrow=1000, ncol=3), matrix(rnorm(1000*3,mean=-0.5,sd=1), nrow=1000, ncol=3)) # 2) calculate the significance according to SVD # using "fdr" significance fdr <- dSVDsignif(data, signif="fdr", num.permutation=10)
Start at 2017-03-27 19:58:37 First, singular value decomposition of the input matrix (with 1000 rows and 9 columns)... Second, determinate the eigens... via automatically deciding on the number of dominant eigens under the cutoff of 1.00e-02 pvalue number of the eigens in consideration: 1 Third, construct the gene-specific projection vector,and calculate distance statistics... Finally, obtain gene significance (fdr) based on 10 permutations... doing row-wise permutations... estimating fdr... 1 (out of 1000) at 2017-03-27 19:58:39 100 (out of 1000) at 2017-03-27 19:58:39 200 (out of 1000) at 2017-03-27 19:58:39 300 (out of 1000) at 2017-03-27 19:58:39 400 (out of 1000) at 2017-03-27 19:58:39 500 (out of 1000) at 2017-03-27 19:58:39 600 (out of 1000) at 2017-03-27 19:58:39 700 (out of 1000) at 2017-03-27 19:58:39 800 (out of 1000) at 2017-03-27 19:58:39 900 (out of 1000) at 2017-03-27 19:58:40 1000 (out of 1000) at 2017-03-27 19:58:40 using stepup procedure... Finish at 2017-03-27 19:58:40 Runtime in total is: 3 secs
# using "pval" significance pval <- dSVDsignif(data, signif="pval", num.permutation=10)
Start at 2017-03-27 19:58:40 First, singular value decomposition of the input matrix (with 1000 rows and 9 columns)... Second, determinate the eigens... via automatically deciding on the number of dominant eigens under the cutoff of 1.00e-02 pvalue number of the eigens in consideration: 1 Third, construct the gene-specific projection vector,and calculate distance statistics... Finally, obtain gene significance (p-value) based on 10 of row-wise permutations... Finish at 2017-03-27 19:58:41 Runtime in total is: 1 secs

Source code

dSVDsignif.r

Source man

dSVDsignif.Rd dSVDsignif.pdf

See also

dFDRscore