st (version 1.2.7)

lait.stat: Correlation-Predicted t-Statistic

Description

lait.stat, laicat.fun, and lai.tscore compute the ``correlation-predicted'' t-statistic of Lai (2008).

Usage

lait.stat(X, L, f=0.2, verbose=TRUE)
lait.fun(L, f=0.2, verbose=TRUE)
lai.tscore(gene, tscore, corr, f=0.2, plot=FALSE)

Arguments

X

data matrix. Note that the columns correspond to variables (``genes'') and the rows to samples.

L

vector with class labels for the two groups.

verbose

print out some (more or less useful) information during computation.

f

smoother span used in lowess (default value: 0.2)

gene

the gene for which the Lai t-score is computed

tscore

a vector with t-scores

corr

a matrix containing correlations

plot

show scatter plot correlations versus t-scores with predicted t-score

Value

lait.stat returns a vector containing correlation-predicted t-statistic for each variable/gene.

The corresponding lait.fun functions return a function that computes the correlation-shared t-statistic when applied to a data matrix (this is very useful for simulations).

The function lai.tscore allows to compute the correlation-predicted t-statistic for a gene given a correlation matrix and a vector of t-statistics.

Details

The correlation-predicted t-statistic for a gene is the t-score predicted by local linear regression using all other genes. For mathematical details see Lai (2008).

References

Lai, Y.. 2008. Genome-wide co-expression based prediction of differential expression. Bioinformatics 24:666-673.

See Also

shrinkcat.stat, cst.stat.

Examples

# NOT RUN {
# load st library 
library("st")

# prostate data set
data(singh2002)
X = singh2002$x
L = singh2002$y

dim(X)      # 102 6033 
length(L)   # 102

# compute correlation-predicted t-score for various choices
# of smoothing span 

# }
# NOT RUN {
score1 = lait.stat(X, L, f=0.1)
idx1 = order(abs(score1), decreasing=TRUE)
idx1[1:10]
# 1072  297 1130 4495 4523 4041 1089  955  373 3848

score3 = lait.stat(X, L, f=0.3)
idx3 = order(abs(score3), decreasing=TRUE)
idx3[1:10]
# 1130  962 1688 1223  583 1118  955  297  698 1219

score5 = lait.stat(X, L, f=0.5)
idx5 = order(abs(score5), decreasing=TRUE)
idx5[1:10]
#  698  962 1223 1219  739 1172  583  694 3785 3370 

score7 = lait.stat(X, L, f=0.7)
idx7 = order(abs(score7), decreasing=TRUE)
idx7[1:10]
#  698  739 1219  962 3785  725  694  735 3370 1172


# pick the one with highest correlation to Student t score
t = studentt.stat(X, L)
cor(t, score1, method="spearman") # 0.4265832
cor(t, score3, method="spearman") # 0.471273
cor(t, score5, method="spearman") # 0.4750564
cor(t, score7, method="spearman") # 0.4666669

# focus on gene 19
t = studentt.stat(X, L)
R = cor(centroids(X, L, lambda.var=0, centered.data=TRUE,
              verbose=TRUE)$centered.data)

lai.tscore(gene=19, t, R, f=0.5, plot=TRUE)

# }