| Title: | The Iterative Bayesian Model Averaging (BMA) algorithm |
|---|---|
| Description: | The iterative Bayesian Model Averaging (BMA) algorithm is a variable selection and classification algorithm with an application of classifying 2-class microarray samples, as described in Yeung, Bumgarner and Raftery (Bioinformatics 2005, 21: 2394-2402). |
| Authors: | Ka Yee Yeung, University of Washington, Seattle, WA, with contributions from Adrian Raftery and Ian Painter |
| Maintainer: | Ka Yee Yeung <[email protected]> |
| License: | GPL (>= 2) |
| Version: | 1.63.0 |
| Built: | 2024-07-14 05:18:08 UTC |
| Source: | https://github.com/bioc/iterativeBMA |
The iterative Bayesian Model Averaging (BMA) algorithm is a variable selection and classification algorithm with an application of classifying 2-class microarray samples, as described in Yeung, Bumgarner and Raftery (Bioinformatics 2005, 21: 2394-2402).
| Package: | iterativeBMA |
| Type: | Package |
| Version: | 0.1.0 |
| Date: | 2005-12-30 |
| License: | GPL version 2 or higher |
The function iterateBMAglm.train selects relevant variables by
iteratively applying the bic.glm function from the BMA package.
The data is assumed to consist of two classes.
The function iterateBMAglm.train.predict combines the training
and prediction phases, and returns the predicted posterior probabilities
that each test sample belongs to class 1.
The function iterateBMAglm.train.predict.test combines the training,
prediction and test phases, and returns a list consisting of the
numbers of selected genes and models using the training data, the number
of classification errors and the Brier Score on the test set.
Ka Yee Yeung, University of Washington, Seattle, WA, with contributions from Adrian Raftery and Ian Painter
Maintainer: Ka Yee Yeung <[email protected]>
Yeung, K.Y., Bumgarner, R.E. and Raftery, A.E. (2005) Bayesian Model Averaging: Development of an improved multi-class, gene selection and classification tool for microarray data. Bioinformatics 21: 2394-2402.
iterateBMAglm.train.predict,
iterateBMAglm.train.predict.test,
bma.predict,
brier.score
library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) ## training phase: select relevant genes ret.bic.glm <- iterateBMAglm.train (train.expr.set=trainData, trainClass, p=100) ## get the selected genes with probne0 > 0 ret.gene.names <- ret.bic.glm$namesx[ret.bic.glm$probne0 > 0] data (testData) ## get the subset of test data with the genes from the last iteration of bic.glm curr.test.dat <- t(exprs(testData)[ret.gene.names,]) ## to compute the predicted probabilities for the test samples y.pred.test <- apply (curr.test.dat, 1, bma.predict, postprobArr=ret.bic.glm$postprob, mleArr=ret.bic.glm$mle) ## compute the Brier Score if the class labels of the test samples are known data (testClass) brier.score (y.pred.test, testClass)library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) ## training phase: select relevant genes ret.bic.glm <- iterateBMAglm.train (train.expr.set=trainData, trainClass, p=100) ## get the selected genes with probne0 > 0 ret.gene.names <- ret.bic.glm$namesx[ret.bic.glm$probne0 > 0] data (testData) ## get the subset of test data with the genes from the last iteration of bic.glm curr.test.dat <- t(exprs(testData)[ret.gene.names,]) ## to compute the predicted probabilities for the test samples y.pred.test <- apply (curr.test.dat, 1, bma.predict, postprobArr=ret.bic.glm$postprob, mleArr=ret.bic.glm$mle) ## compute the Brier Score if the class labels of the test samples are known data (testClass) brier.score (y.pred.test, testClass)
This function computes the predicted posterior probability that each test sample belongs to class 1. It assumes 2-class data, and requires the true class labels to be known.
bma.predict (newdataArr, postprobArr, mleArr)bma.predict (newdataArr, postprobArr, mleArr)
newdataArr |
a vector consisting of the data from a test sample. |
postprobArr |
a vector consisting of the posterior probability of each BMA selected model. |
mleArr |
matrix with one row per model and one column per variable giving the maximum likelihood estimate of each coefficient for each BMA selected model. |
Let Y be the response variable (class labels for samples in our case). In Bayesian Model Averaging (BMA), the posterior probability of Y=1 given the training set is the weighted average of the posterior probability of Y=1 given the training set and model M multiplied by the posterior probability of model M given the training set, summing over a set of models M.
A real number between zero and one, representing the predicted posterior probability.
Raftery, A.E. (1995). Bayesian model selection in social research (with Discussion). Sociological Methodology 1995 (Peter V. Marsden, ed.), pp. 111-196, Cambridge, Mass.: Blackwells.
Yeung, K.Y., Bumgarner, R.E. and Raftery, A.E. (2005) Bayesian Model Averaging: Development of an improved multi-class, gene selection and classification tool for microarray data. Bioinformatics 21: 2394-2402.
brier.score,
iterateBMAglm.train
library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) ## training phase: select relevant genes ret.bic.glm <- iterateBMAglm.train (train.expr.set=trainData, trainClass, p=100) ## get the selected genes with probne0 > 0 ret.gene.names <- ret.bic.glm$namesx[ret.bic.glm$probne0 > 0] data (testData) ## get the subset of test data with the genes from the last iteration of bic.glm curr.test.dat <- t(exprs(testData)[ret.gene.names,]) ## to compute the predicted probabilities for the test samples y.pred.test <- apply (curr.test.dat, 1, bma.predict, postprobArr=ret.bic.glm$postprob, mleArr=ret.bic.glm$mle) ## compute the Brier Score if the class labels of the test samples are known data (testClass) brier.score (y.pred.test, testClass)library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) ## training phase: select relevant genes ret.bic.glm <- iterateBMAglm.train (train.expr.set=trainData, trainClass, p=100) ## get the selected genes with probne0 > 0 ret.gene.names <- ret.bic.glm$namesx[ret.bic.glm$probne0 > 0] data (testData) ## get the subset of test data with the genes from the last iteration of bic.glm curr.test.dat <- t(exprs(testData)[ret.gene.names,]) ## to compute the predicted probabilities for the test samples y.pred.test <- apply (curr.test.dat, 1, bma.predict, postprobArr=ret.bic.glm$postprob, mleArr=ret.bic.glm$mle) ## compute the Brier Score if the class labels of the test samples are known data (testClass) brier.score (y.pred.test, testClass)
The Brier Score is a probabilistic number of errors that takes the predicted probabilities into consideration. A small Brier Score indicates high prediction accuracy. This function assumes 2-class data, and requires the true class labels to be known.
brier.score (predictedArr, truthArr)brier.score (predictedArr, truthArr)
predictedArr |
a vector consisting of the predicted probabilities that the test sample belongs to class 1. |
truthArr |
a zero-one vector indicating the known class labels of
the test samples. We assume this vector has the same
length as |
The Brier Score computes the sum of squares of the differences between the true class and the predicted probability over all test samples. If the predicted probabilities are constrained to equal to 0 or 1, the Brier Score is equal to the total number of classification errors.
A non-negative real number.
Brier, G.W. (1950) Verification of forecasts expressed in terms of probability. Monthly Weather Review 78: 1-3.
Yeung, K.Y., Bumgarner, R.E. and Raftery, A.E. (2005) Bayesian Model Averaging: Development of an improved multi-class, gene selection and classification tool for microarray data. Bioinformatics 21: 2394-2402.
bma.predict,
iterateBMAglm.train.predict
library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) data (testData) ret.vec <- iterateBMAglm.train.predict (train.expr.set=trainData, test.expr.set=testData, trainClass, p=100) ## compute the Brier Score data (testClass) brier.score (ret.vec, testClass)library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) data (testData) ret.vec <- iterateBMAglm.train.predict (train.expr.set=trainData, test.expr.set=testData, trainClass, p=100) ## compute the Brier Score data (testClass) brier.score (ret.vec, testClass)
This is a univariate technique to select relevant genes in classification of microarray data. In classifying samples of microarray data, this ratio is computed for each gene. A large between-groups to within-groups sum-of-squares ratio indicates a potentially relevant gene.
BssWssFast (X, givenClassArr, numClass = 2)BssWssFast (X, givenClassArr, numClass = 2)
X |
data matrix where columns are variables and rows are observations. In the case of gene expression data, the columns (variables) represent genes, while the rows (observations) represent samples or experiments. |
givenClassArr |
class vector for the observations (samples or experiments). Class numbers are assumed to start from 0, and the length of this class vector should be equal to the number of rows in X. In the case of 2-class data, we expect the class vector consists of zero's and one's. |
numClass |
number of classes. The default is 2. |
This function is called by iterateBMAglm.2class.
A list of 2 elements are returned:
x |
A vector containing the BSS/WSS ratios in descending order. |
ix |
A vector containing the indices corresponding to the sorted ratios. |
Dudoit, S., Fridlyand, J. and Speed, T.P. (2002) Comparison of discrimination methods for the classification of tumors using gene expression data. Journal of the American Statistical Association 97: 77-87.
Yeung, K.Y., Bumgarner, R.E. and Raftery, A.E. (2005) Bayesian Model Averaging: Development of an improved multi-class, gene selection and classification tool for microarray data. Bioinformatics 21: 2394-2402.
iterateBMAglm.train, trainData,
trainClass
data(trainData) data(trainClass) ret.bsswss <- BssWssFast (X=t(exprs(trainData)), givenClassArr=trainClass, numClass = 2)data(trainData) data(trainClass) ret.bsswss <- BssWssFast (X=t(exprs(trainData)), givenClassArr=trainClass, numClass = 2)
Create a visualization of the models and variables selected by the iterative BMA algorithm.
imageplot.iterate.bma (bicreg.out, color="default", ...)imageplot.iterate.bma (bicreg.out, color="default", ...)
bicreg.out |
An object of type 'bicreg', 'bic.glm' or 'bic.surv' |
color |
The color of the plot. The value "default" uses the current default R color scheme for image. The value "blackandwhite" produces a black and white image. |
... |
Other parameters to be passed to the image and axis functions. |
This function is a modification of the imageplot.bma
function from the BMA package. The difference is that
variables (genes) with probne0 equal to 0 are removed
before plotting. The arguments of this function is identical
to those in imageplot.bma.
An heatmap-style image, with the BMA selected variables on the vertical
axis, and the BMA selected models on the horizontal axis. The variables
(genes) are sorted in descreasing order of the posterior probability
that the variable is not equal to 0 (probne0) from top to
bottom. The models are sorted in descreasing order of the
model posterior probability (postprob) from left to right.
The BMA and Biobase packages are required.
Clyde, M. (1999) Bayesian Model Averaging and Model Search Strategies (with discussion). In Bayesian Statistics 6. J.M. Bernardo, A.P. Dawid, J.O. Berger, and A.F.M. Smith eds. Oxford University Press, pages 157-185.
Yeung, K.Y., Bumgarner, R.E. and Raftery, A.E. (2005) Bayesian Model Averaging: Development of an improved multi-class, gene selection and classification tool for microarray data. Bioinformatics 21: 2394-2402.
library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) ## training phase: select relevant genes ret.bic.glm <- iterateBMAglm.train (train.expr.set=trainData, trainClass, p=100) ## produce an image plot to visualize the selected genes and models imageplot.iterate.bma (ret.bic.glm)library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) ## training phase: select relevant genes ret.bic.glm <- iterateBMAglm.train (train.expr.set=trainData, trainClass, p=100) ## produce an image plot to visualize the selected genes and models imageplot.iterate.bma (ret.bic.glm)
Classification and variable selection on microarray data. This is a multivariate technique to select a small number of relevant variables (typically genes) to classify microarray samples. This function performs the training phase. The data is assumed to consist of two classes. Logistic regression is used for classification.
iterateBMAglm.train (train.expr.set, train.class, p=100, nbest=10, maxNvar=30, maxIter=20000, thresProbne0=1)iterateBMAglm.train (train.expr.set, train.class, p=100, nbest=10, maxNvar=30, maxIter=20000, thresProbne0=1)
train.expr.set |
an |
train.class |
class vector for the observations (samples or experiments) in the training data. Class numbers are assumed to start from 0, and the length of this class vector should be equal to the number of rows in train.dat. Since we assume 2-class data, we expect the class vector consists of zero's and one's. |
p |
a number indicating the maximum number of top univariate genes used in the iterative BMA algorithm. This number is assumed to be less than the total number of genes in the training data. A larger p usually requires longer computational time as more iterations of the BMA algorithm are potentially applied. The default is 100. |
nbest |
a number specifying the number of models of each size
returned to |
maxNvar |
a number indicating the maximum number of variables used in
each iteration of |
maxIter |
a number indicating the maximum of iterations of
|
thresProbne0 |
a number specifying the threshold for the posterior
probability that each variable (gene) is non-zero (in
percent). Variables (genes) with such posterior
probability less than this threshold are dropped in
the iterative application of |
The training phase consists of first
ordering all the variables (genes) by a univariate measure
called between-groups to within-groups sums-of-squares (BSS/WSS)
ratio, and then iteratively applying the bic.glm algorithm
from the BMA package. In the first application of
the bic.glm algorithm, the top maxNvar univariate
ranked genes are used. After each application of the bic.glm
algorithm, the genes with probne0 < thresProbne0
are dropped, and the next univariate ordered genes are added
to the BMA window.
An object of class bic.glm returned by the last iteration
of bic.glm. The object is a list consisting of
the following components:
namesx |
the names of the variables in the last iteration of
|
postprob |
the posterior probabilities of the models selected. |
deviance |
the estimated model deviances. |
label |
labels identifying the models selected. |
bic |
values of BIC for the models. |
size |
the number of independent variables in each of the models. |
which |
a logical matrix with one row per model and one column per variable indicating whether that variable is in the model. |
probne0 |
the posterior probability that each variable is non-zero (in percent). |
postmean |
the posterior mean of each coefficient (from model averaging). |
postsd |
the posterior standard deviation of each coefficient (from model averaging). |
condpostmean |
the posterior mean of each coefficient conditional on the variable being included in the model. |
condpostsd |
the posterior standard deviation of each coefficient conditional on the variable being included in the model. |
mle |
matrix with one row per model and one column per variable giving the maximum likelihood estimate of each coefficient for each model. |
se |
matrix with one row per model and one column per variable giving the standard error of each coefficient for each model. |
reduced |
a logical indicating whether any variables were dropped before model averaging. |
dropped |
a vector containing the names of those variables dropped before model averaging. |
call |
the matched call that created the bma.lm object. |
The BMA and Biobase packages are required.
Raftery, A.E. (1995). Bayesian model selection in social research (with Discussion). Sociological Methodology 1995 (Peter V. Marsden, ed.), pp. 111-196, Cambridge, Mass.: Blackwells.
Yeung, K.Y., Bumgarner, R.E. and Raftery, A.E. (2005) Bayesian Model Averaging: Development of an improved multi-class, gene selection and classification tool for microarray data. Bioinformatics 21: 2394-2402.
iterateBMAglm.train.predict,
iterateBMAglm.train.predict.test,
bma.predict,
brier.score
library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) ## training phase: select relevant genes ret.bic.glm <- iterateBMAglm.train (train.expr.set=trainData, trainClass, p=100) ## get the selected genes with probne0 > 0 ret.gene.names <- ret.bic.glm$namesx[ret.bic.glm$probne0 > 0] ## show the posterior probabilities of selected models ret.bic.glm$postprob data (testData) ## get the subset of test data with the genes from the last iteration of bic.glm curr.test.dat <- t(exprs(testData)[ret.gene.names,]) ## to compute the predicted probabilities for the test samples y.pred.test <- apply (curr.test.dat, 1, bma.predict, postprobArr=ret.bic.glm$postprob, mleArr=ret.bic.glm$mle) ## compute the Brier Score if the class labels of the test samples are known data (testClass) brier.score (y.pred.test, testClass)library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) ## training phase: select relevant genes ret.bic.glm <- iterateBMAglm.train (train.expr.set=trainData, trainClass, p=100) ## get the selected genes with probne0 > 0 ret.gene.names <- ret.bic.glm$namesx[ret.bic.glm$probne0 > 0] ## show the posterior probabilities of selected models ret.bic.glm$postprob data (testData) ## get the subset of test data with the genes from the last iteration of bic.glm curr.test.dat <- t(exprs(testData)[ret.gene.names,]) ## to compute the predicted probabilities for the test samples y.pred.test <- apply (curr.test.dat, 1, bma.predict, postprobArr=ret.bic.glm$postprob, mleArr=ret.bic.glm$mle) ## compute the Brier Score if the class labels of the test samples are known data (testClass) brier.score (y.pred.test, testClass)
Classification and variable selection on microarray data. This is a multivariate technique to select a small number of relevant variables (typically genes) to classify microarray samples. This function performs the training, and prediction steps. The data is assumed to consist of two classes. Logistic regression is used for classification.
iterateBMAglm.train.predict (train.expr.set, test.expr.set, train.class, p=100, nbest=10, maxNvar=30, maxIter=20000, thresProbne0=1)iterateBMAglm.train.predict (train.expr.set, test.expr.set, train.class, p=100, nbest=10, maxNvar=30, maxIter=20000, thresProbne0=1)
train.expr.set |
an |
test.expr.set |
an |
train.class |
class vector for the observations (samples or experiments) in the training data. Class numbers are assumed to start from 0, and the length of this class vector should be equal to the number of rows in train.dat. Since we assume 2-class data, we expect the class vector consists of zero's and one's. |
p |
a number indicating the maximum number of top univariate genes used in the iterative BMA algorithm. This number is assumed to be less than the total number of genes in the training data. A larger p usually requires longer computational time as more iterations of the BMA algorithm are potentially applied. The default is 100. |
nbest |
a number specifying the number of models of each size
returned to |
maxNvar |
a number indicating the maximum number of variables used in
each iteration of |
maxIter |
a number indicating the maximum of iterations of
|
thresProbne0 |
a number specifying the threshold for the posterior
probability that each variable (gene) is non-zero (in
percent). Variables (genes) with such posterior
probability less than this threshold are dropped in
the iterative application of |
This function consists of the training phase and the prediction
phase. The training phase consists of first
ordering all the variables (genes) by a univariate measure
called between-groups to within-groups sums-of-squares (BSS/WSS)
ratio, and then iteratively applying the bic.glm algorithm
from the BMA package. The prediction phase uses the variables
(genes) selected in the training phase to classify the samples
in the test set.
A vector consisting of the predicted probability that each test sample belongs to class 1 is returned.
The BMA and Biobase packages are required.
Raftery, A.E. (1995). Bayesian model selection in social research (with Discussion). Sociological Methodology 1995 (Peter V. Marsden, ed.), pp. 111-196, Cambridge, Mass.: Blackwells.
Yeung, K.Y., Bumgarner, R.E. and Raftery, A.E. (2005) Bayesian Model Averaging: Development of an improved multi-class, gene selection and classification tool for microarray data. Bioinformatics 21: 2394-2402.
iterateBMAglm.train,
iterateBMAglm.train.predict.test,
brier.score
library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) data (testData) ret.vec <- iterateBMAglm.train.predict (train.expr.set=trainData, test.expr.set=testData, trainClass, p=100) ## compute the Brier Score data (testClass) brier.score (ret.vec, testClass)library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) data (testData) ret.vec <- iterateBMAglm.train.predict (train.expr.set=trainData, test.expr.set=testData, trainClass, p=100) ## compute the Brier Score data (testClass) brier.score (ret.vec, testClass)
Classification and variable selection on microarray data. This is a multivariate technique to select a small number of relevant variables (typically genes) to classify microarray samples. This function performs the training, prediction and testing steps. The data is assumed to consist of two classes, and the classes of the test data is assumed to be known. Logistic regression is used for classification.
iterateBMAglm.train.predict.test (train.expr.set, test.expr.set, train.class, test.class, p=100, nbest=10, maxNvar=30, maxIter=20000, thresProbne0=1)iterateBMAglm.train.predict.test (train.expr.set, test.expr.set, train.class, test.class, p=100, nbest=10, maxNvar=30, maxIter=20000, thresProbne0=1)
train.expr.set |
an |
test.expr.set |
an |
train.class |
class vector for the observations (samples or experiments) in the training data. Class numbers are assumed to start from 0, and the length of this class vector should be equal to the number of rows in train.dat. Since we assume 2-class data, we expect the class vector consists of zero's and one's. |
test.class |
class vector for the observations (samples or experiments) in the test data. Class numbers are assumed to start from 0, and the length of this class vector should be equal to the number of rows in test.dat. Since we assume 2-class data, we expect the class vector consists of zero's and one's. |
p |
a number indicating the maximum number of top univariate genes used in the iterative BMA algorithm. This number is assumed to be less than the total number of genes in the training data. A larger p usually requires longer computational time as more iterations of the BMA algorithm are potentially applied. The default is 100. |
nbest |
a number specifying the number of models of each size
returned to |
maxNvar |
a number indicating the maximum number of variables used in
each iteration of |
maxIter |
a number indicating the maximum of iterations of
|
thresProbne0 |
a number specifying the threshold for the posterior
probability that each variable (gene) is non-zero (in
percent). Variables (genes) with such posterior
probability less than this threshold are dropped in
the iterative application of |
This function consists of the training phase, prediction phase,
and the testing phase. The training phase consists of first
ordering all the variables (genes) by a univariate measure
called between-groups to within-groups sums-of-squares (BSS/WSS)
ratio, and then iteratively applying the bic.glm algorithm
from the BMA package. The prediction phase uses the variables
(genes) selected in the training phase to classify the samples
in the test set. The testing phase assumes that the class labels
of the samples in the test set are known, and computes the number of
classification errors and the Brier Score.
A list consisting of 4 elements are returned:
num.genes |
The number of relevant genes selected using the training data. |
num.model |
The number of models selected using the training data. |
num.err |
The number of classification errors produced when the the predicted class labels of the test samples are compared to the known class labels. |
brierScore |
The Brier Score computed using the predicted and known class labels of the test samples. The Brier Score represents a probabilistic number of errors. A small Brier Score implies high prediction accuracy. |
The BMA and Biobase packages are required.
Raftery, A.E. (1995). Bayesian model selection in social research (with Discussion). Sociological Methodology 1995 (Peter V. Marsden, ed.), pp. 111-196, Cambridge, Mass.: Blackwells.
Yeung, K.Y., Bumgarner, R.E. and Raftery, A.E. (2005) Bayesian Model Averaging: Development of an improved multi-class, gene selection and classification tool for microarray data. Bioinformatics 21: 2394-2402.
iterateBMAglm.train,
iterateBMAglm.train.predict
library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) data (testData) data (testClass) iterateBMAglm.train.predict.test (train.expr.set=trainData, test.expr.set=testData, trainClass, testClass, p=100)library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) data (testData) data (testClass) iterateBMAglm.train.predict.test (train.expr.set=trainData, test.expr.set=testData, trainClass, testClass, p=100)
This function repeatedly calls bic.glm from the
BMA package until all variables are exhausted.
The data is assumed to consist of
two classes. Logistic regression is used for classification.
iterateBMAglm.wrapper (sortedA, y, nbest=10, maxNvar=30, maxIter=20000, thresProbne0=1)iterateBMAglm.wrapper (sortedA, y, nbest=10, maxNvar=30, maxIter=20000, thresProbne0=1)
sortedA |
data matrix where columns are variables and rows are observations. The variables (columns) are assumed to be sorted using a univariate measure. In the case of gene expression data, the columns (variables) represent genes, while the rows (observations) represent samples or experiments. |
y |
class vector for the observations (samples or experiments) in the training data. Class numbers are assumed to start from 0, and the length of this class vector should be equal to the number of rows in sortedA. Since we assume 2-class data, we expect the class vector consists of zero's and one's. |
nbest |
a number specifying the number of models of each size
returned to |
maxNvar |
a number indicating the maximum number of variables used in
each iteration of |
maxIter |
a number indicating the maximum of iterations of
|
thresProbne0 |
a number specifying the threshold for the posterior
probability that each variable (gene) is non-zero (in
percent). Variables (genes) with such posterior
probability less than this threshold are dropped in
the iterative application of |
In this function, the variables are assumed to be sorted, and
bic.glm is called repeatedly. In the first application of
the bic.glm algorithm, the top maxNvar univariate
ranked genes are used. After each application of the bic.glm
algorithm, the genes with probne0 < thresProbne0
are dropped, and the next univariate ordered genes are added
to the BMA window.
The function iterateBMAglm.train calls BssWssFast before
calling this function.
Using this function, users can experiment with alternative
univariate measures.
If all variables are exhausted, an object of class
bic.glm returned by the last iteration
of bic.glm. Otherwise, -1 is returned.
The object of class bic.glm is a list consisting
of the following components:
namesx |
the names of the variables in the last iteration of
|
postprob |
the posterior probabilities of the models selected. |
deviance |
the estimated model deviances. |
label |
labels identifying the models selected. |
bic |
values of BIC for the models. |
size |
the number of independent variables in each of the models. |
which |
a logical matrix with one row per model and one column per variable indicating whether that variable is in the model. |
probne0 |
the posterior probability that each variable is non-zero (in percent). |
postmean |
the posterior mean of each coefficient (from model averaging). |
postsd |
the posterior standard deviation of each coefficient (from model averaging). |
condpostmean |
the posterior mean of each coefficient conditional on the variable being included in the model. |
condpostsd |
the posterior standard deviation of each coefficient conditional on the variable being included in the model. |
mle |
matrix with one row per model and one column per variable giving the maximum likelihood estimate of each coefficient for each model. |
se |
matrix with one row per model and one column per variable giving the standard error of each coefficient for each model. |
reduced |
a logical indicating whether any variables were dropped before model averaging. |
dropped |
a vector containing the names of those variables dropped before model averaging. |
call |
the matched call that created the bma.lm object. |
The BMA and Biobase packages are required.
Raftery, A.E. (1995). Bayesian model selection in social research (with Discussion). Sociological Methodology 1995 (Peter V. Marsden, ed.), pp. 111-196, Cambridge, Mass.: Blackwells.
Yeung, K.Y., Bumgarner, R.E. and Raftery, A.E. (2005) Bayesian Model Averaging: Development of an improved multi-class, gene selection and classification tool for microarray data. Bioinformatics 21: 2394-2402.
iterateBMAglm.train,
iterateBMAglm.train.predict,
iterateBMAglm.train.predict.test,
BssWssFast
library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) ## Use the BSS/WSS ratio to rank all genes in the training data sorted.vec <- BssWssFast (t(exprs(trainData)), trainClass, numClass = 2) ## get the top ranked 50 genes sorted.train.dat <- t(exprs(trainData[sorted.vec$ix[1:50], ])) ## run iterative bic.glm ret.bic.glm <- iterateBMAglm.wrapper (sorted.train.dat, y=trainClass) ## The above commands are equivalent to the following ret.bic.glm <- iterateBMAglm.train (train.expr.set=trainData, trainClass, p=50)library (Biobase) library (BMA) library (iterativeBMA) data(trainData) data(trainClass) ## Use the BSS/WSS ratio to rank all genes in the training data sorted.vec <- BssWssFast (t(exprs(trainData)), trainClass, numClass = 2) ## get the top ranked 50 genes sorted.train.dat <- t(exprs(trainData[sorted.vec$ix[1:50], ])) ## run iterative bic.glm ret.bic.glm <- iterateBMAglm.wrapper (sorted.train.dat, y=trainClass) ## The above commands are equivalent to the following ret.bic.glm <- iterateBMAglm.train (train.expr.set=trainData, trainClass, p=50)
This is an adapted leukemia (ALL, AML) dataset from Golub et al. (1999). This is a zero-one vector for 34 test samples. Class 0 represents an ALL sample, while class 1 represents an AML sample.
data(testClass)data(testClass)
The vector is called testClass.
The golubEsets bioconductor data package, or
http://www.broad.mit.edu/cgi-bin/cancer/datasets.cgi.
Golub, T.R., Slonim, D.K., Tamayo, P., Huard, C., Gaasenbeek, M., Mesirov, J.P., Coller, H., Loh, M.L., Downing, J.R., Caligiuri, M.A., et al. (1999) Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 286: 531-7.
This is an adapted leukemia (ALL, AML) ExpressionSet from Golub et al. (1999).
This ExpressionSet consists of the expression levels from 38 ALL or AML samples
(rows), and 100 genes (columns). This dataset is used
as an example test data in our examples.
data(testData)data(testData)
The ExpressionSet is called testData. Each entry
in the matrix represents the expression level of one gene from an ALL
or AML sample.
For illustration purposes, a subset of 100 genes from the package golubEsets
is included in this package.
The golubEsets bioconductor data package, or
http://www.broad.mit.edu/cgi-bin/cancer/datasets.cgi.
Golub, T.R., Slonim, D.K., Tamayo, P., Huard, C., Gaasenbeek, M., Mesirov, J.P., Coller, H., Loh, M.L., Downing, J.R., Caligiuri, M.A., et al. (1999) Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 286: 531-7.
This is an adapted leukemia (ALL, AML) dataset from Golub et al. (1999). This is a zero-one vector for 38 training samples. Class 0 represents an ALL sample, while class 1 represents an AML sample.
data(trainClass)data(trainClass)
The vector is called trainClass.
The golubEsets bioconductor data package, or
http://www.broad.mit.edu/cgi-bin/cancer/datasets.cgi.
Golub, T.R., Slonim, D.K., Tamayo, P., Huard, C., Gaasenbeek, M., Mesirov, J.P., Coller, H., Loh, M.L., Downing, J.R., Caligiuri, M.A., et al. (1999) Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 286: 531-7.
This is an adapted leukemia (ALL, AML) ExpressionSet from Golub et al. (1999).
This ExpressionSet consists of the expression levels from 38 ALL or AML samples
(rows), and 100 genes (columns). This dataset is used
as an example training data in our examples.
data(trainData)data(trainData)
The ExpressionSet is called trainData. Each entry
in the exprs matrix represents the expression level of one gene from an ALL
or AML sample.
For illustration purposes, a subset of 100 genes from the package golubEsets
is included in this package.
The golubEsets bioconductor data package, or
http://www.broad.mit.edu/cgi-bin/cancer/datasets.cgi.
Golub, T.R., Slonim, D.K., Tamayo, P., Huard, C., Gaasenbeek, M., Mesirov, J.P., Coller, H., Loh, M.L., Downing, J.R., Caligiuri, M.A., et al. (1999) Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 286: 531-7.