Package 'superpc'

Title: Supervised Principal Components
Description: Does prediction in the case of a censored survival outcome, or a regression outcome, using the "supervised principal component" approach. 'Superpc' is especially useful for high-dimensional data when the number of features p dominates the number of samples n (p >> n paradigm), as generated, for instance, by high-throughput technologies.
Authors: Eric Bair [aut], Jean-Eudes Dazard [cre, ctb], Rob Tibshirani [ctb]
Maintainer: Jean-Eudes Dazard <[email protected]>
License: GPL (>= 3) | file LICENSE
Version: 1.12
Built: 2024-10-02 02:41:06 UTC
Source: https://github.com/jedazard/superpc

Help Index

Cross-validation for supervised principal components

Description

This function uses a form of cross-validation to estimate the optimal feature threshold in supervised principal components

Usage

superpc.cv(fit,
               data, 
               n.threshold=20,
               n.fold=NULL,
               folds=NULL,
               n.components=3, 
               min.features=5, 
               max.features=nrow(data$x),
               compute.fullcv= TRUE,
               compute.preval=TRUE, 
               xl.mode=c("regular","firsttime","onetime","lasttime"), 
               xl.time=NULL,
               xl.prevfit=NULL)

Arguments

fit

Object returned by superpc.train

data

Data object of form described in superpc.train documentation

n.threshold

Number of thresholds to consider. Default 20.
n.fold

Number of cross-validation folds. default is around 10 (program pick a convenient value based on the sample size

folds

List of indices of cross-validation folds (optional)

n.components

Number of cross-validation components to use: 1,2 or 3.

min.features

Minimum number of features to include in determining range for threshold. Default 5.
max.features

Maximum number of features to include in determining range for threshold. Default is total number of features in the dataset

compute.fullcv

Should full cross-validation be done?

compute.preval

Should full pre-validation be done?

xl.mode

Used by Excel interface only
xl.time

Used by Excel interface only
xl.prevfit

Used by Excel interface only

Details

This function uses a form of cross-validation to estimate the optimal feature threshold in supervised principal components. To avoid prolems with fitting Cox models to samll validation datastes, it uses the "pre-validation" approach of Tibshirani and Efron (2002)

Value

threshold

Vector of thresholds considered
nonzero

Number of features exceeding each value of the threshold
scor.preval

Likelihood ratio scores from pre-validation
scor

Full CV scores
folds

Indices of CV folds used
featurescores.folds

Feature scores for each fold
v.preval

The pre-validated predictors
type

problem type
call

calling sequence

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

## Not run: 
set.seed(332)

#generate some data
x <- matrix(rnorm(50*30), ncol=30)
y <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
censoring.status <- sample(c(rep(1,20), rep(0,10)))

featurenames <- paste("feature", as.character(1:50), sep="")
data <- list(x=x, 
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)

a <- superpc.train(data, type="survival")
aa <- superpc.cv(a, data)

## End(Not run)

Decorrelate features with respect to competing predictors

Description

Fits a linear model to the features as a function of some competing predictors. Replaces the features by the residual from this fit. These "decorrelated" features are then used in the superpc model building process, to explicitly look for predictors that are independent of the competing predictors. Useful for example, when the competing predictors are clinical predictors like stage, grade etc.

Usage

superpc.decorrelate(x, 
                        competing.predictors)

Arguments

x

matrix of features. Different features in different rows, one observation per column
competing.predictors

List of one or more competing predictors. Discrete predictors should be factors

Value

Returns lm (linear model) fit of rows of x on compeiting predictors.

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

set.seed(332)

#generate some data
x <- matrix(rnorm(50*30), ncol=30)
y <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
censoring.status <- sample(c(rep(1,20), rep(0,10)))

featurenames <- paste("feature", as.character(1:50), sep="")
competing.predictors <- list(pred1=rnorm(30), 
                             pred2=as.factor(sample(c(1,2), 
                                             replace=TRUE, 
                                             size=30)))

#decorrelate x. Remember to decorrelate test data in the same way, before making predictions.
foo <- superpc.decorrelate(x, competing.predictors)
xnew <- t(foo$res)

#now use xnew in superpc
data <- list(x=xnew, 
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)
a <- superpc.train(data, type="survival")

#etc.

Fit predictive model using outcome of supervised principal components

Description

Fit predictive model using outcome of supervised principal components, via either coxph (for surival data) or lm (for regression data)

Usage

superpc.fit.to.outcome(fit, 
                           data.test, 
                           score, 
                           competing.predictors=NULL, 
                           print=TRUE, 
                           iter.max=5)

Arguments

fit

Object returned by superpc.train.
data.test

Data object for prediction. Same form as data object documented in superpc.train.
score

Supervised principal component score, from superpc.predict.
competing.predictors

Optional - a list of competing predictors to be included in the model.
print

Should a summary of the fit be printed? Default TRUE.
iter.max

Max number of iterations used in predictive model fit. Default 5. Currently only relevant for Cox PH model.

Value

Returns summary of coxph or lm fit.

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

set.seed(332)

#generate some data
x <- matrix(rnorm(50*30), ncol=30)
y <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
ytest <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
censoring.status <- sample(c(rep(1,20), rep(0,10)))
censoring.status.test <- sample(c(rep(1,20), rep(0,10)))

featurenames <- paste("feature", as.character(1:50), sep="")
data <- list(x=x, 
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)
data.test <- list(x=x, 
                  y=ytest, 
                  censoring.status=censoring.status.test, 
                  featurenames=featurenames)

a <- superpc.train(data, type="survival")
fit <- superpc.predict(a, 
                       data, 
                       data.test, 
                       threshold=1.0, 
                       n.components=1, 
                       prediction.type="continuous")
superpc.fit.to.outcome(a, 
                       data, 
                       fit$v.pred)

Return a list of the important predictors

Description

Return a list of the important predictor

Usage

superpc.listfeatures(data, 
                         train.obj, 
                         fit.red, 
                         fitred.cv=NULL,
                         num.features=NULL, 
                         component.number=1)

Arguments

data

Data object

train.obj

Object returned by superpc.train
fit.red

Object returned by superpc.predict.red, applied to training set
fitred.cv

(Optional) object returned by superpc.predict.red.cv
num.features

Number of features to list. Default is all features.
component.number

Number of principal component (1,2, or 3) used to determine feature importance scores

Value

Returns matrix of features and their importance scores, in order of decreasing absolute value of importance score. The importance score is the correlation of the reduced predictor and the full supervised PC predictor. It also lists the raw score- for survival data, this is the Cox score for that feature; for regression, it is the standardized regression coefficient. If fitred.cv is supplied, the function also reports the average rank of the gene in the cross-validation folds, and the proportion of times that the gene is chosen (at the given threshold) in the cross-validation folds.

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

set.seed(332)

#generate some data
x <- matrix(rnorm(50*30), ncol=30)
y <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
ytest <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
censoring.status <- sample(c(rep(1,20), rep(0,10)))
censoring.status.test <- sample(c(rep(1,20), rep(0,10)))

featurenames <- paste("feature", as.character(1:50), sep="")
data <- list(x=x, 
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)
data.test <- list(x=x, 
                  y=ytest, 
                  censoring.status=censoring.status.test, 
                  featurenames=featurenames)

a <- superpc.train(data, type="survival")
fit.red <- superpc.predict.red(a, 
                               data, 
                               data.test, 
                               .6)
superpc.listfeatures(data, 
                     a,  
                     fit.red, 
                     num.features=10)

Compute values of likelihood ratio test from supervised principal components fit

Description

Compute values of likelihood ratio test from supervised principal components fit

Usage

superpc.lrtest.curv(object, 
                        data, 
                        newdata, 
                        n.components=1, 
                        threshold=NULL, 
                        n.threshold=20)

Arguments

object

Object returned by superpc.train.
data

List of training data, of form described in superpc.train documentation.
newdata

List of test data; same form as training data.
n.components

Number of principal components to compute. Should be 1,2 or 3.
threshold

Set of thresholds for scores; default is n.threshold values equally spaced over the range of the feature scores.
n.threshold

Number of thresholds to use; default 20. Should be 1,2 or 3.

Value

lrtest

Values of likelihood ratio test statistic
comp2

Description of 'comp2'
threshold

Thresholds used
num.features

Number of features exceeding threshold
type

Type of outcome variable
call

calling sequence

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

set.seed(332)

#generate some data
x <- matrix(rnorm(50*30), ncol=30)
y <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
ytest <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
censoring.status <- sample(c(rep(1,20), rep(0,10)))
censoring.status.test <- sample(c(rep(1,20), rep(0,10)))

featurenames <- paste("feature", as.character(1:50), sep="")
data <- list(x=x, 
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)
data.test <- list(x=x, 
                  y=ytest, 
                  censoring.status=censoring.status.test, 
                  featurenames=featurenames)

a <- superpc.train(data, type="survival")
aa <- superpc.lrtest.curv(a, data, data.test)
#superpc.plot.lrtest(aa)

Display the superpc Package News

Description

Function to display the log file NEWS of updates of the superpc package.

Usage

superpc.news(...)

Arguments

...

Further arguments passed to or from other methods.

Value

None.

Note

End-user function.

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Plot likelhiood ratio test statistics

Description

Plot likelhiood ratio test statistics from output of superpc.predict

Usage

superpc.plot.lrtest(object.lrtestcurv, 
                        call.win.metafile=FALSE)

Arguments

object.lrtestcurv

Output from superpc.lrtest.curv
call.win.metafile

For use by PAM Excel interface

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

set.seed(332)

#generate some data
x <- matrix(rnorm(50*30), ncol=30)
y <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
ytest <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
censoring.status <- sample(c(rep(1,20), rep(0,10)))
censoring.status.test <- sample(c(rep(1,20), rep(0,10)))

featurenames <- paste("feature", as.character(1:50), sep="")
data <- list(x=x, 
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)
data.test <- list(x=x, 
                  y=ytest, 
                  censoring.status=censoring.status.test, 
                  featurenames=featurenames)

a <- superpc.train(data, type="survival")
bb <- superpc.lrtest.curv(a, 
                          data, 
                          data.test)
superpc.plot.lrtest(bb)

Plot output from superpc.cv

Description

Plots pre-validation results from plotcv, to aid in choosing best threshold

Usage

superpc.plotcv(object, 
                   cv.type=c("full","preval"),
                   smooth=TRUE, 
                   smooth.df=10, 
                   call.win.metafile=FALSE, ...)

Arguments

object

Object returned by superpc.cv.
cv.type

Type of cross-validation used - "full" (Default; this is "standard" cross-validation; recommended) and "preval"- pre-validation.
smooth

Should plot be smoothed? Only relevant to "preval". Default FALSE.
smooth.df

Degrees of freedom for smooth.spline, default 10. If NULL, then degrees of freedom is estimated by cross-validation.
call.win.metafile

Ignore: for use by PAM Excel program.
...

Additional plotting args to be passed to matplot.

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

## Not run: 
set.seed(332)

#generate some data
x <- matrix(rnorm(50*30), ncol=30)
y <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
censoring.status <- sample(c(rep(1,20), rep(0,10)))

featurenames <- paste("feature", as.character(1:50), sep="")
data <- list(x=x, 
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)

a <- superpc.train(data, type="survival")
aa <- superpc.cv(a,data)

superpc.plotcv(aa)

## End(Not run)

Plot likelihood ratio test statistics from supervised principal components predictor

Description

Plot likelihood ratio test statistics from supervised principal components predictor

Usage

superpc.plotred.lrtest(object.lrtestred, 
                           call.win.metafile=FALSE)

Arguments

object.lrtestred

Output from either superpc.predict.red or superpc.predict.redcv
call.win.metafile

Used only by PAM Excel interface call to function

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

## Not run: 
set.seed(332)

#generate some data
x <- matrix(rnorm(50*30), ncol=30)
y <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
ytest <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
censoring.status <- sample(c(rep(1,20), rep(0,10)))
censoring.status.test <- sample(c(rep(1,20), rep(0,10)))

featurenames <- paste("feature", as.character(1:50), sep="")
data <- list(x=x, 
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)
data.test <- list(x=x, 
                  y=ytest, 
                  censoring.status=censoring.status.test, 
                  featurenames=featurenames)

a <- superpc.train(data, type="survival")
aa <- superpc.cv(a, data)
fit.red <- superpc.predict.red(a, 
                               data, 
                               data.test, 
                               .6)
fit.redcv <- superpc.predict.red.cv(fit.red, 
                                    aa, 
                                    data, 
                                    .6)
superpc.plotred.lrtest(fit.redcv)

## End(Not run)

Form principal components predictor from a trained superpc object

Description

Computes supervised principal components, using scores from "object"

Usage

superpc.predict(object, 
                    data, 
                    newdata, 
                    threshold, 
                    n.components=3, 
                    prediction.type=c("continuous","discrete","nonzero"), 
                    n.class=2)

Arguments

object

Obect returned by superpc.train

data

List of training data, of form described in superpc.train documentation,

newdata

List of test data; same form as training data

threshold

Threshold for scores: features with abs(score) > threshold are retained.

n.components

Number of principal components to compute. Should be 1,2 or 3.

prediction.type

"continuous" for raw principal component(s); "discrete" for principal component categorized in equal bins; "nonzero" for indices of features that pass the threshold

n.class

Number of classes into which predictor is binned (for prediction.type="discrete"

Value

v.pred

Supervised principal componients predictor
u

U matrix from svd of feature matrix x
d

singual values from svd of feature matrix x
which.features

Indices of features exceeding threshold
n.components

Number of supervised principal components requested
call

calling sequence

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

set.seed(332)

#generate some data
x <- matrix(rnorm(50*30), ncol=30)
y <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
ytest <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
censoring.status <- sample(c(rep(1,20), rep(0,10)))
censoring.status.test <- sample(c(rep(1,20), rep(0,10)))

featurenames <- paste("feature", as.character(1:50), sep="")
data <- list(x=x, 
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)
data.test <- list(x=x, 
                  y=ytest, 
                  censoring.status=censoring.status.test, 
                  featurenames=featurenames)

a <- superpc.train(data, type="survival")
fit <- superpc.predict(a, 
                       data, 
                       data.test, 
                       threshold=1.0, 
                       n.components=1)
plot(fit$v.pred, ytest)

Feature selection for supervised principal components

Description

Forms reduced models to approximate the supervised principal component predictor.

Usage

superpc.predict.red(fit, 
                        data, 
                        data.test, 
                        threshold, 
                        n.components=3, 
                        n.shrinkage=20, 
                        shrinkages=NULL,
                        compute.lrtest=TRUE,
                        sign.wt="both",
                        prediction.type=c("continuous", "discrete"), 
                        n.class=2)

Arguments

fit

Object returned by superpc.train
data

Training data object, of form described in superpc.train dcoumentation

data.test

Test data object; same form as train

threshold

Feature score threshold; usually estimated from superpc.cv
n.components

Number of principal components to examine; should equal 1,2, etc up to the number of components used in training

n.shrinkage

Number of shrinkage values to consider. Default 20.

shrinkages

Shrinkage values to consider. Default NULL.

compute.lrtest

Should the likelihood ratio test be computed? Default TRUE

sign.wt

Signs of feature weights allowed: "both", "pos", or "neg"
prediction.type

Type of prediction: "continuous" (Default) or "discrete". In the latter, superprc score is divided into n.class groups
n.class

Number of groups for discrete predictor. Default 2.

Details

Soft-thresholding by each of the "shrinkages" values is applied to the PC loadings. This reduce the number of features used in the model. The reduced predictor is then used in place of the supervised PC predictor.

Value

shrinkages

Shrinkage values used

lrtest.reduced

Likelihood ratio tests for reduced models

num.features

Number of features used in each reduced model

feature.list

List of features used in each reduced model

coef

Least squares coefficients for each reduced model
import

Importance scores for features
wt

Weight for each feature, in constructing the reduced predictor
v.test

Outcome predictor from reduced models. Array of n.shrinkage by (number of test observations)

v.test.1df

Outcome combined predictor from reduced models. Array of n.shrinkage by (number of test observations)

n.components

Number of principal components used
type

Type of outcome
call

calling sequence

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

set.seed(332)

#generate some data
x <- matrix(rnorm(50*30), ncol=30)
y <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
ytest <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
censoring.status <- sample(c(rep(1,20), rep(0,10)))
censoring.status.test <- sample(c(rep(1,20), rep(0,10)))

featurenames <- paste("feature", as.character(1:50), sep="")
data <- list(x=x,
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)
data.test <- list(x=x, 
                  y=ytest, 
                  censoring.status=censoring.status.test, 
                  featurenames=featurenames)

a <- superpc.train(data, type="survival")
fit.red <- superpc.predict.red(a,
                               data, 
                               data.test, 
                               threshold=.6)
superpc.plotred.lrtest(fit.red)

Cross-validation of feature selection for supervised principal components

Description

Applies superpc.predict.red to cross-validation folds generates in superpc.cv. Uses the output to evaluate reduced models, and compare them to the full supervised principal components predictor.

Usage

superpc.predict.red.cv(fitred, 
                           fitcv, 
                           data, 
                           threshold, 
                           sign.wt="both")

Arguments

fitred

Output of superpc.predict.red
fitcv

Output of superpc.cv

data

Training data object

threshold

Feature score threshold; usually estimated from superpc.cv
sign.wt

Signs of feature weights allowed: "both", "pos", or "neg"

Value

lrtest.reduced

Likelihood ratio tests for reduced models

components

Number of supervised principal components used
v.preval.red

Outcome predictor from reduced models. Array of num.reduced.models by (number of test observations)
type

Type of outcome
call

calling sequence

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

## Not run: 
set.seed(332)

#generate some data
x <- matrix(rnorm(50*20), ncol=20)
y <- 10 + svd(x[1:10,])$v[,1] + .1*rnorm(20)
ytest <- 10 + svd(x[1:10,])$v[,1] + .1*rnorm(20)
censoring.status <- sample(c(rep(1,15), rep(0,5)))
censoring.status.test <- sample(c(rep(1,15), rep(0,5)))

featurenames <- paste("feature", as.character(1:50), sep="")
data <- list(x=x, 
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)
data.test <- list(x=x,
                  y=ytest, 
                  censoring.status=censoring.status.test, 
                  featurenames=featurenames)

a <- superpc.train(data, type="survival")
aa <- superpc.cv(a, data)
fit.red <- superpc.predict.red(a,
                               data, 
                               data.test, 
                               threshold=.6)
fit.redcv <- superpc.predict.red.cv(fit.red, 
                                    aa, 
                                    data, 
                                    threshold=.6)

## End(Not run)

Plot outcome predictions from superpc

Description

Plots outcome predictions from superpc

Usage

superpc.predictionplot(train.obj, 
                           data, 
                           data.test,
                           threshold, 
                           n.components=3,
                           n.class=2, 
                           shrinkage=NULL, 
                           call.win.metafile=FALSE)

Arguments

train.obj

Object returned by superpc.train

data

List of training data, of form described in superpc.train documentation
data.test

List of test data; same form as training data

threshold

Threshold for scores: features with abs(score) > threshold are retained.

n.components

Number of principal components to compute. Should be 1,2 or 3.

n.class

Number of classes for survival stratification. Only applicable for survival data. Default 2.
shrinkage

Shrinkage to be applied to feature loadings. Default is NULL, meaning no shrinkage
call.win.metafile

Used only by Excel interface call to function

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

set.seed(332)

#generate some data
x <- matrix(rnorm(50*30), ncol=30)
y <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
ytest <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
censoring.status <- sample(c(rep(1,20), rep(0,10)))
censoring.status.test <- sample(c(rep(1,20), rep(0,10)))

featurenames <- paste("feature", as.character(1:50), sep="")
data <- list(x=x, 
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)
data.test <- list(x=x, 
                  y=ytest, 
                  censoring.status=censoring.status.test, 
                  featurenames=featurenames)

a <- superpc.train(data, type="survival")
superpc.predictionplot(a, 
                       data, 
                       data.test, 
                       threshold=1)

Make rainbow plot of superpc and compeiting predictors

Description

Makes a heatmap display of outcome predictions from superpc, along with expected survival time, and values of competing predictors.

Usage

superpc.rainbowplot(data, 
                        pred, 
                        sample.labels,
                        competing.predictors,
                        call.win.metafile=FALSE)

Arguments

data

List of (test) data, of form described in superpc.train documentation
pred

Superpc score from superpc.predict or superpc.predict.red
sample.labels

Vector of sample labels of test data
competing.predictors

List of competing predictors to be plotted
call.win.metafile

Used only by Excel interface call to function

Details

Any censored survival times are estimated by E(T|T > C), where $C$ is the observed censoring time and the Kaplan-Meier estimate from the training set is used to estimate the expectation.

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

set.seed(332)

#generate some data
x <- matrix(rnorm(50*30), ncol=30)
y <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
ytest <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
censoring.status <- sample(c(rep(1,20), rep(0,10)))
censoring.status.test <- sample(c(rep(1,20), rep(0,10)))

featurenames <- paste("feature", as.character(1:50), sep="")
competing.predictors.test <- list(pred1=rnorm(30), 
                                  pred2=as.factor(sample(c(1,2),
                                                  replace=TRUE,
                                                  size=30)))
                                                  
data <- list(x=x, 
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)
data.test <- list(x=x, 
                  y=ytest, 
                  censoring.status=censoring.status.test, 
                  featurenames=featurenames)
sample.labels <- paste("te", as.character(1:20), sep="")

a <- superpc.train(data, type="survival")
pred <- superpc.predict(a, 
                        data, 
                        data.test, 
                        threshold=.25, 
                        n.components=1)$v.pred
superpc.rainbowplot(data, 
                    pred, 
                    sample.labels, 
                    competing.predictors=competing.predictors.test)

Prediction by supervised principal components

Description

Does prediction of a quantitative regression or survival outcome, by the supervised principal components method.

Usage

superpc.train(data, 
                  type=c("survival", "regression"), 
                  s0.perc=NULL)

Arguments

data

Data object with components x- p by n matrix of features, one observation per column; y- n-vector of outcome measurements; censoring.status- n-vector of censoring censoring.status (1= died or event occurred, 0=survived, or event was censored), needed for a censored survival outcome
type

Problem type: "survival" for censored survival outcome, or "regression" for simple quantitative outcome
s0.perc

Factor for denominator of score statistic, between 0 and 1: the percentile of standard deviation values added to the denominator. Default is 0.5 (the median)

Details

Compute wald scores for each feature (gene), for later use in superpc.predict and superpc.cv

Value

feature.scores

Score for each feature (gene)
type

problem type
s0.perc

Factor for denominator of score statistic
call

calling sequence

Author(s)

  • "Eric Bair, Ph.D."

  • "Jean-Eudes Dazard, Ph.D."

  • "Rob Tibshirani, Ph.D."

Maintainer: "Jean-Eudes Dazard, Ph.D."

References

  • E. Bair and R. Tibshirani (2004). "Semi-supervised methods to predict patient survival from gene expression data." PLoS Biol, 2(4):e108.

  • E. Bair, T. Hastie, D. Paul, and R. Tibshirani (2006). "Prediction by supervised principal components." J. Am. Stat. Assoc., 101(473):119-137.

Examples

set.seed(332)

#generate some data
x <- matrix(rnorm(50*30), ncol=30)
y <- 10 + svd(x[1:50,])$v[,1] + .1*rnorm(30)
censoring.status <- sample(c(rep(1,20), rep(0,10)))

featurenames <- paste("feature", as.character(1:50), sep="")
data <- list(x=x, 
             y=y, 
             censoring.status=censoring.status, 
             featurenames=featurenames)

a <- superpc.train(data, type="survival")