| Title: | A Biomarker Validation Approach for Classification and Predicting Survival Using Metabolomics Signature |
|---|---|
| Description: | An approach to identifies metabolic biomarker signature for metabolic data by discovering predictive metabolite for predicting survival and classifying patients into risk groups. Classifiers are constructed as a linear combination of predictive/important metabolites, prognostic factors and treatment effects if necessary. Several methods were implemented to reduce the metabolomics matrix such as the principle component analysis of Wold Svante et al. (1987) <doi:10.1016/0169-7439(87)80084-9> , the LASSO method by Robert Tibshirani (1998) <doi:10.1002/(SICI)1097-0258(19970228)16:4%3C385::AID-SIM380%3E3.0.CO;2-3>, the elastic net approach by Hui Zou and Trevor Hastie (2005) <doi:10.1111/j.1467-9868.2005.00503.x>. Sensitivity analysis on the quantile used for the classification can also be accessed to check the deviation of the classification group based on the quantile specified. Large scale cross validation can be performed in order to investigate the mostly selected predictive metabolites and for internal validation. During the evaluation process, validation is accessed using the hazard ratios (HR) distribution of the test set and inference is mainly based on resampling and permutations technique. |
| Authors: | Olajumoke Evangelina Owokotomo [aut, cre], Ziv Shkedy [aut] |
| Maintainer: | Olajumoke Evangelina Owokotomo <[email protected]> |
| License: | GPL-3 |
| Version: | 1.1.0 |
| Built: | 2025-02-14 04:27:17 UTC |
| Source: | https://github.com/olajumokeevangelina/metabolicsurv |
The function does cross validation for Lasso, Elastic net and Ridge regressions models before the survial analysis and classification. The survival analysis is based on the selected metabolites in the presence or absene of prognostic factors.
CVLasoelacox(Survival, Censor, Mdata, Prognostic, Quantile = 0.5, Metlist = NULL, Standardize = TRUE, Reduce = TRUE, Select = 15, Alpha = 1, Fold = 4, Ncv = 10, nlambda = 100)CVLasoelacox(Survival, Censor, Mdata, Prognostic, Quantile = 0.5, Metlist = NULL, Standardize = TRUE, Reduce = TRUE, Select = 15, Alpha = 1, Fold = 4, Ncv = 10, nlambda = 100)
Survival |
A vector of survival time with length equals to number of subjects |
Censor |
A vector of censoring indicator |
Mdata |
A large or small metabolic profile matrix. A matrix with metabolic profiles where the number of rows should be equal to the number of metabolites and number of columns should be equal to number of patients. |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
Quantile |
The cut off value for the classifier, default is the median cutoff |
Metlist |
A list of metabolites to be considered in the model usually smaller than the metabolites in the Mdata . Default is to use all metabolites available and it is advisable to be greater than 17. |
Standardize |
A Logical flag for the standardization of the metabolite matrix, prior to fitting the model sequence. The coefficients are always returned on the original scale. Default is standardize=TRUE. |
Reduce |
A boolean parameter indicating if the metabolic profile matrix should be reduced, default is TRUE and larger metabolic profile matrix is reduced by supervised pca approach and first pca is extracted from the reduced matrix to be used in the classifier. |
Select |
Number of metabolites (default is 15) to be selected from supervised PCA. This is valid only if the argument Reduce=TRUE |
Alpha |
The mixing parameter for glmnet (see |
Fold |
number of folds to be used for the cross validation. Its value ranges between 3 and the numbe rof subjects in the dataset |
Ncv |
Number of validations to be carried out. The default is 25. |
nlambda |
The number of lambda values - default is 100 as in glmnet. |
The function performs the cross validations for Lasso, Elastic net and Ridge regressions models for Cox proportional hazard model. Metabolites are selected at each iteration and then use for the classifier. Which implies that predictive metabolites signature is varied from one cross validation to the other depending on selection. The underline idea is to investigate the Hazard Ratio for the train and test data based on the optimal lambda selected for the non-zero shrinkage coefficients, the nonzero selected metabolites will thus be used in the survival analysis and in calculation of the risk scores for each sets of data.
A object of class cvle is returned with the following values
Coef.mat |
A matrix of coefficients with rows equals to number of cross validations and columns equals to number of metabolites. |
Runtime |
A vector of runtime for each iteration measured in seconds. |
lambda |
A vector of estimated optimum lambda for each iterations. |
n |
A vector of the number of selected metabolites |
HRTrain |
A matrix of survival information for the training dataset. It has three columns representing the estimated HR, the 95% lower confidence interval and the 95% upper confidence interval. |
HRTest |
A matrix of survival information for the test dataset. It has three columns representing the estimated HR, the 95% lower confidence interval and the 95% upper confidence interval. |
pld |
A vector of partial likelihood deviance at each cross validations. |
Met.mat |
A matrix with 0 and 1. Number of rows equals to number of iterations and number of columns equals to number of metabolites. 1 indicates that the particular metabolite was selected or had nonzero coefficient and otherwise it is zero. |
Mdata |
The Metabolite data matrix that was used for the analysis either same as Mdata or a reduced version. |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
coxph,
EstimateHR, glmnet, Lasoelacox
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Results = CVLasoelacox(Survival = Data$Survival,Censor = Data$Censor, Mdata = t(Data$Mdata),Prognostic = Data$Prognostic, Quantile = 0.5, Metlist = NULL,Standardize = TRUE, Reduce=FALSE, Select=15, Alpha = 1,Fold = 4,Ncv = 10,nlambda = 100) ## NUMBER OF SELECTED METABOLITES PER CV Results@n ## GET THE MATRIX OF COEFFICIENTS [email protected] ## SURVIVAL INFORMATION OF THE TRAIN DATASET Results@HRTrain ## SURVIVAL INFORMATION OF THE TEST DATASET Results@HRTest## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Results = CVLasoelacox(Survival = Data$Survival,Censor = Data$Censor, Mdata = t(Data$Mdata),Prognostic = Data$Prognostic, Quantile = 0.5, Metlist = NULL,Standardize = TRUE, Reduce=FALSE, Select=15, Alpha = 1,Fold = 4,Ncv = 10,nlambda = 100) ## NUMBER OF SELECTED METABOLITES PER CV Results@n ## GET THE MATRIX OF COEFFICIENTS Results@Coef.mat ## SURVIVAL INFORMATION OF THE TRAIN DATASET Results@HRTrain ## SURVIVAL INFORMATION OF THE TEST DATASET Results@HRTest
Class of object returned by function CVLasoelacox.
## S4 method for signature 'cvle' show(object) ## S4 method for signature 'cvle' summary(object) ## S4 method for signature 'cvle,missing' plot(x, y, type = 1, ...)## S4 method for signature 'cvle' show(object) ## S4 method for signature 'cvle' summary(object) ## S4 method for signature 'cvle,missing' plot(x, y, type = 1, ...)
object |
A cvle class object |
x |
A cvle class object |
y |
missing |
type |
Plot type. 1 distribution of the HR under training and test set. 2 HR vs number selected metabolites. |
... |
The usual extra arguments to generic functions — see |
Coef.matA matrix of coefficients with rows equals to number of cross validations and columns equals to number of metabolites.
RuntimeA vector of runtime for each iteration measured in seconds.
lambdaA vector of estimated optimum lambda for each iterations.
nA vector of the number of selected metabolites
Met.matA matrix with 0 and 1. Number of rows equals to number of iterations and number of columns equals to number of metabolites. 1 indicates that the particular metabolite was selected or had nonzero coefficient and otherwise it is zero.
HRTrainA matrix of survival information for the training dataset. It has three columns representing the estimated HR, the 95% lower confidence interval and the 95% upper confidence interval.
HRTestA matrix of survival information for the test dataset. It has three columns representing the estimated HR, the 95% lower confidence interval and the 95% upper confidence interval.
pldA vector of partial likelihood deviance at each cross validations.
MdataThe metabolite matrix that was used for the analysis which can either be the full the full data or a reduced supervised PCA version.
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
EstimateHR, glmnet, Lasoelacox
## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## USE THE FUNCTION Eg = CVLasoelacox(Survival = Data$Survival,Censor = Data$Censor, Mdata = t(Data$Mdata),Prognostic = Data$Prognostic, Quantile = 0.5, Metlist = NULL,Standardize = TRUE, Reduce=FALSE, Select=15, Alpha = 1,Fold = 4,Ncv = 10,nlambda = 100) ## GET THE CLASS OF THE OBJECT class(Eg) # An "cvle" Class ## METHOD THAT CAN BE USED FOR THIS CLASS show(Eg) summary(Eg) plot(Eg, type =3)## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## USE THE FUNCTION Eg = CVLasoelacox(Survival = Data$Survival,Censor = Data$Censor, Mdata = t(Data$Mdata),Prognostic = Data$Prognostic, Quantile = 0.5, Metlist = NULL,Standardize = TRUE, Reduce=FALSE, Select=15, Alpha = 1,Fold = 4,Ncv = 10,nlambda = 100) ## GET THE CLASS OF THE OBJECT class(Eg) # An "cvle" Class ## METHOD THAT CAN BE USED FOR THIS CLASS show(Eg) summary(Eg) plot(Eg, type =3)
This function does cross validation for the Majority votes based classification.
CVMajorityvotes(Survival, Censor, Prognostic = NULL, Mdata, Reduce = TRUE, Select = 15, Fold = 3, Ncv = 100)CVMajorityvotes(Survival, Censor, Prognostic = NULL, Mdata, Reduce = TRUE, Select = 15, Fold = 3, Ncv = 100)
Survival |
A vector of survival time with length equals to number of subjects |
Censor |
A vector of censoring indicator |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
Mdata |
A large or small metabolic profile matrix. A matrix with metabolic profiles where the number of rows should be equal to the number of metabolites and number of columns should be equal to number of patients. |
Reduce |
A boolean parameter indicating if the metabolic profile matrix should be reduced, default is TRUE and larger metabolic profile matrix is reduced by supervised pca approach and first pca is extracted from the reduced matrix to be used in the classifier. |
Select |
Number of metabolites (default is 15) to be selected from supervised PCA. This is valid only if th argument Reduce=TRUE |
Fold |
Number of times in which the dataset is divided. Default is 3 which implies dataset will be divided into three groups and 2/3 of the dataset will be the train datset and 1/3 will be to train the results. |
Ncv |
The Number of cross validation loop. Default is 50 but it is recommended to have at least 100. |
This function does cross validation for the Majority votes based classification which is a cross validated approach to Majorityvotes.
A object of class cvmv is returned with the following values
HRTrain |
A matrix of survival information for the training dataset. It has three columns representing the estimated HR, the 95% lower confidence interval and the 95% upper confidence interval. |
HRTest |
A matrix of survival information for the test dataset. It has three columns representing the estimated HR, the 95% lower confidence interval and the 95% upper confidence interval. |
Ncv |
The number of cross validation used |
Mdata |
The Metabolite data matrix that was used for the analysis either same as Mdata or a reduced version. |
Progfact |
The names of prognostic factors used |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Result = CVMajorityvotes(Survival=Data$Survival,Censor=Data$Censor, Prognostic=Data$Prognostic, Mdata=t(Data$Mdata), Reduce=FALSE, Select=15, Fold=3, Ncv=10) ## GET THE CLASS OF THE OBJECT class(Result) # An "cvmv" Class ## METHOD THAT CAN BE USED FOR THE RESULT show(Result) summary(Result) plot(Result)## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Result = CVMajorityvotes(Survival=Data$Survival,Censor=Data$Censor, Prognostic=Data$Prognostic, Mdata=t(Data$Mdata), Reduce=FALSE, Select=15, Fold=3, Ncv=10) ## GET THE CLASS OF THE OBJECT class(Result) # An "cvmv" Class ## METHOD THAT CAN BE USED FOR THE RESULT show(Result) summary(Result) plot(Result)
The function performs cross validation for each metabolite depending the number of fold which guides the division into the train and testing dataset. The classifier is then obtained on the training dataset to be validated on the test dataset
CVMetSpecificCoxPh(Fold = 3, Survival, Mdata, Censor, Reduce = TRUE, Select = 150, Prognostic = NULL, Quantile = 0.5, Ncv = 3)CVMetSpecificCoxPh(Fold = 3, Survival, Mdata, Censor, Reduce = TRUE, Select = 150, Prognostic = NULL, Quantile = 0.5, Ncv = 3)
Fold |
Number of times in which the dataset is divided. Default is 3 which implies dataset will be divided into three groups and 2/3 of the dataset will be the train datset and 1/3 will be to train the results. |
Survival |
A vector of survival time with length equals to number of subjects |
Mdata |
A large or small metabolic profile matrix. A matrix with metabolic profiles where the number of rows should be equal to the number of metabolites and number of columns should be equal to number of patients. |
Censor |
A vector of censoring indicator |
Reduce |
A boolean parameter indicating if the metabolic profile matrix should be reduced, default is TRUE and larger metabolic profile matrix is reduced by supervised pca approach and first pca is extracted from the reduced matrix to be used in the classifier. |
Select |
Number of metabolites (default is 15) to be selected from supervised PCA. This is valid only if th argument Reduce=TRUE |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
Quantile |
The cut off value for the classifier, default is the median cutoff |
Ncv |
The Number of cross validation loop. Default is 50 but it is recommended to have at least 100. |
This function performs the cross validation for metabolite by metabolite analysis. The data will firstly be divided into data train dataset and test datset. Furthermore, a metabolite-specific model is fitted on train data and a classifier is built. In addition, the classifier is then evaluated on test dataset for each particular metabolite. The Process is repeated for all the full or reduced metabolites to obtaind the HR statistics of the low risk group. The following steps depends on the number of cross validation specified.
A object of class cvmm is returned with the following values
HRTrain |
The Train dataset HR statistics for each metabolite by the number of CV |
HRTest |
The Test dataset HR statistics for each metabolite by the number of CV |
train |
The selected subjects for each CV in the train dataset |
train |
The selected subjects for each CV in the test dataset |
n.mets |
The number of metabolite used in the analysis |
Ncv |
The number of cross validation performed |
Rdata |
The Metabolite data matrix that was used for the analysis either same as Mdata or a reduced version. |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
coxph,
EstimateHR, MSpecificCoxPh,
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Result = CVMetSpecificCoxPh(Fold=3,Survival=Data$Survival, Mdata=t(Data$Mdata),Censor= Data$Censor,Reduce=TRUE, Select=150,Prognostic=Data$Prognostic,Quantile = 0.5,Ncv=3) ## GET THE CLASS OF THE OBJECT class(Result) # An "cvmm" Class ## METHOD THAT CAN BE USED FOR THE RESULT show(Result) summary(Result) plot(Result)## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Result = CVMetSpecificCoxPh(Fold=3,Survival=Data$Survival, Mdata=t(Data$Mdata),Censor= Data$Censor,Reduce=TRUE, Select=150,Prognostic=Data$Prognostic,Quantile = 0.5,Ncv=3) ## GET THE CLASS OF THE OBJECT class(Result) # An "cvmm" Class ## METHOD THAT CAN BE USED FOR THE RESULT show(Result) summary(Result) plot(Result)
Class of object returned by function CVMetSpecificCoxPh.
## S4 method for signature 'cvmm' show(object) ## S4 method for signature 'cvmm' summary(object, which = 1) ## S4 method for signature 'cvmm,ANY' plot(x, y, which = 1, ...)## S4 method for signature 'cvmm' show(object) ## S4 method for signature 'cvmm' summary(object, which = 1) ## S4 method for signature 'cvmm,ANY' plot(x, y, which = 1, ...)
object |
A CVMetSpecificCoxPh class object |
which |
This specify which metabolite for which estimated HR information need to be visualized. By default results of the first metabolite is used. |
x |
A CVMetSpecificCoxPh class object |
y |
missing |
... |
The usual extra arguments to generic functions — see |
plot signature(x = "cvmm"): Plots for CVMetSpecificCoxPh class analysis results.
Any parameters of plot.default may be passed on to this particular plot method.
HRTrainA 3-way array, The first dimension is the number of metabolites, the second dimension is the HR statistics for the low risk group in the train dataset (HR,1/HR LCI, UCI) while the third dimension is the number of cross validation performed.
HRTestA 3-way array, The first dimension is the number of metabolites, the second dimension is the HR statistics for the low risk group in the test dataset (HR,1/HR LCI, UCI) while the third dimension is the number of cross validation performed.
trainThe selected subjects for each CV in the train dataset
testThe selected subjects for each CV in the test dataset
n.metsThe number of metabolite used in the analysis
NcvThe number of cross validation performed
RdataThe Metabolite data matrix that was used for the analysis either same as Mdata or a reduced version
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## USING THE FUNCTION Result = CVMetSpecificCoxPh(Fold=3,Survival=Data$Survival, Mdata=t(Data$Mdata),Censor= Data$Censor,Reduce=TRUE, Select=150,Prognostic=Data$Prognostic,Quantile = 0.5,Ncv=3) ## GET THE CLASS OF THE OBJECT class(Result) # An "cvmm" Class ## METHOD THAT CAN BE USED FOR THIS CLASS show(Result) summary(Result) plot(Result)## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## USING THE FUNCTION Result = CVMetSpecificCoxPh(Fold=3,Survival=Data$Survival, Mdata=t(Data$Mdata),Censor= Data$Censor,Reduce=TRUE, Select=150,Prognostic=Data$Prognostic,Quantile = 0.5,Ncv=3) ## GET THE CLASS OF THE OBJECT class(Result) # An "cvmm" Class ## METHOD THAT CAN BE USED FOR THIS CLASS show(Result) summary(Result) plot(Result)
Class of object returned by function CVMajorityvotes.
## S4 method for signature 'cvmv' show(object) ## S4 method for signature 'cvmv' summary(object) ## S4 method for signature 'cvmv,ANY' plot(x, y, ...)## S4 method for signature 'cvmv' show(object) ## S4 method for signature 'cvmv' summary(object) ## S4 method for signature 'cvmv,ANY' plot(x, y, ...)
object |
A cvmv class object |
x |
A cvmv class object |
y |
missing |
... |
The usual extra arguments to generic functions — see |
HRTrainA matrix of survival information for the training dataset. It has three columns representing the estimated HR, the 95% lower confidence interval and the 95% upper confidence interval.
HRTestA matrix of survival information for the test dataset. It has three columns representing the estimated HR, the 95% lower confidence interval and the 95% upper confidence interval.
NcvThe number of cross validation used
MdataThe Metabolite data matrix that was used for the analysis either same as Mdata or a reduced version.
ProgfactThe names of prognostic factors used
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
Majorityvotes, CVPcaPls, SurvPcaClass, SurvPlsClass
## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## USING THE FUNCTION Result = CVMajorityvotes(Survival=Data$Survival,Censor=Data$Censor, Prognostic=Data$Prognostic, Mdata=t(Data$Mdata), Reduce=FALSE, Select=15, Fold=3, Ncv=10) ## GET THE CLASS OF THE OBJECT class(Result) # A "cvmv" Class ## METHOD THAT CAN BE USED FOR THE RESULT show(Result) summary(Result) plot(Result)## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## USING THE FUNCTION Result = CVMajorityvotes(Survival=Data$Survival,Censor=Data$Censor, Prognostic=Data$Prognostic, Mdata=t(Data$Mdata), Reduce=FALSE, Select=15, Fold=3, Ncv=10) ## GET THE CLASS OF THE OBJECT class(Result) # A "cvmv" Class ## METHOD THAT CAN BE USED FOR THE RESULT show(Result) summary(Result) plot(Result)
This function does cross validation for the analysis performs by SurvPcaClass and SurvPlsClass functions where the dimension reduction methods can either be PCA and PLS.
CVPcaPls(Fold = 3, Survival, Mdata, Censor, Reduce = TRUE, Select = 15, Prognostic = NULL, Ncv = 5, DR = "PCA")CVPcaPls(Fold = 3, Survival, Mdata, Censor, Reduce = TRUE, Select = 15, Prognostic = NULL, Ncv = 5, DR = "PCA")
Fold |
Number of times in which the dataset is divided. Default is 3 which implies dataset will be divided into three groups and 2/3 of the dataset will be the train datset and 1/3 will be to train the results. |
Survival |
A vector of survival time with length equals to number of subjects |
Mdata |
A large or small metabolic profile matrix. A matrix with metabolic profiles where the number of rows should be equal to the number of metabolites and number of columns should be equal to number of patients. |
Censor |
A vector of censoring indicator |
Reduce |
A boolean parameter indicating if the metabolic profile matrix should be reduced, default is TRUE and larger metabolic profile matrix is reduced by supervised pca approach and first pca is extracted from the reduced matrix to be used in the classifier. |
Select |
Number of metabolites (default is 15) to be selected from supervised PCA. This is valid only if th argument Reduce=TRUE |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
Ncv |
The Number of cross validation loop. Default is 50 but it is recommended to have at least 100. |
DR |
The dimension reduction method. It can be either "PCA" for Principle components analysis or "PLS" for Partial least squares. |
This function does cross validation for the analysis using two reduction method. The reduction method can be PCA or PLS. If it is PCA then the SurvPcaClass is internally used for the cross validation and SurvPlsClass otherwise.
A object of class cvpp is returned with the following values
Result |
A dataframe containg the estimated Hazard ratio of the test dataset and the training dataset |
Ncv |
The number of cross validation performed |
Method |
The dimesion reduction method used |
CVtrain |
The training dataset indices matrix used for the cross validation |
CVtest |
The test dataset indices matrix used for the cross validation |
Select |
The number of metabolite used for the dimesion reduction method used |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
Bair E, Hastie T, Debashis P, Tibshirani R (2006). “Prediction by supervised principal components.” American Statistics Association,, 101(473), 119-137.
Vinzi VE, Chin WW, Henseler J, Wang H (2010). Handbook of Partial Least Squares: Concepts, Methods and Applications, 1st edition. Springer Publishing Company, Incorporated.
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Result = CVPcaPls(Fold = 4, Survival = Data$Survival, Mdata = t(Data$Mdata), Censor = Data$Censor, Reduce=TRUE, Select=19, Prognostic= Data$Prognostic,Ncv=55,DR ="PLS") ## GET THE CLASS OF THE OBJECT class(Result) # An "cvpp" Class ## METHOD THAT CAN BE USED FOR THE RESULT show(Result) summary(Result) plot(Result)## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Result = CVPcaPls(Fold = 4, Survival = Data$Survival, Mdata = t(Data$Mdata), Censor = Data$Censor, Reduce=TRUE, Select=19, Prognostic= Data$Prognostic,Ncv=55,DR ="PLS") ## GET THE CLASS OF THE OBJECT class(Result) # An "cvpp" Class ## METHOD THAT CAN BE USED FOR THE RESULT show(Result) summary(Result) plot(Result)
Class of object returned by function CVPcaPls.
## S4 method for signature 'cvpp' show(object) ## S4 method for signature 'cvpp' summary(object) ## S4 method for signature 'cvpp,missing' plot(x, y, ...)## S4 method for signature 'cvpp' show(object) ## S4 method for signature 'cvpp' summary(object) ## S4 method for signature 'cvpp,missing' plot(x, y, ...)
object |
A cvpp class object |
x |
A cvpp class object |
y |
missing |
... |
The usual extra arguments to generic functions — see |
ResultsA dataframe containg the estimated Hazard ratio of the test dataset and the training dataset
NcvThe number of cross validation performed
MethodThe dimesion reduction method used
CVtrainThe training dataset indices matrix used for the cross validation
CVtestThe test dataset indices matrix used for the cross validation
SelectThe number of metabolite used for the dimesion reduction method used
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
CVPcaPls, SurvPcaClass, SurvPlsClass
## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## USING THE FUNCTION Result = CVPcaPls(Fold = 4, Survival = Data$Survival, Mdata = t(Data$Mdata), Censor = Data$Censor, Reduce=TRUE, Select=19, Prognostic= Data$Prognostic,Ncv=55,DR ="PLS") ## GET THE CLASS OF THE OBJECT class(Result) # A "cvpp" Class ## METHOD THAT CAN BE USED FOR THE RESULT show(Result) summary(Result) plot(Result)## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## USING THE FUNCTION Result = CVPcaPls(Fold = 4, Survival = Data$Survival, Mdata = t(Data$Mdata), Censor = Data$Censor, Reduce=TRUE, Select=19, Prognostic= Data$Prognostic,Ncv=55,DR ="PLS") ## GET THE CLASS OF THE OBJECT class(Result) # A "cvpp" Class ## METHOD THAT CAN BE USED FOR THE RESULT show(Result) summary(Result) plot(Result)
This function does cross validation for the metabolite by metabolite analysis while sequentially increasing the number of metabolites as specified.
CVSim(Object, Top = seq(5, 100, by = 5), Survival, Censor, Prognostic = NULL)CVSim(Object, Top = seq(5, 100, by = 5), Survival, Censor, Prognostic = NULL)
Object |
An object of class |
Top |
The Top k number of metabolites to be used |
Survival |
A vector of survival time with length equals to number of subjects |
Censor |
A vector of censoring indicator |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
The function is a cross validation version of the function SIMet. This function firstly processes the cross validation for the metabolite by metabolite analysis results, and then sequentially considers top k metabolites. The function recompute first PCA or PLS on train data and estimate risk scores on both test and train data only on the metabolite matrix with top k metabolites. Patients are then classified as having low or high risk based on the test data where the cutoff used is median of the risk score. The process is repeated for each top K metabolite sets.
A object of class cvsim is returned with the following values
HRpca |
A 3-way array in which first, second, and third dimensions correspond to number of metabolites, Hazard ratio infromation(Estimated HR, LowerCI and UpperCI), and number of cross validation respectively. This contains the estimated HR on test data and dimension reduction method is PCA. |
HRpls |
A 3-way array in which first, second, and third dimensions correspond to number of metabolites, Hazard ratio infromation(Estimated HR, LowerCI and UpperCI), and number of cross validation respectively. This contains the estimated HR on test data and dimension reduction method is PLS. |
Nmets |
The number of metabolites in the reduced matrix |
Ncv |
The number of cross validation done |
Top |
A sequence of top k metabolites considered. Default is Top=seq(5,100,by=5) |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## GETTING THE cvmm OBJECT Result = CVMetSpecificCoxPh(Fold=3,Survival=Data$Survival, Mdata=t(Data$Mdata),Censor= Data$Censor,Reduce=TRUE,Select=150, Prognostic=Data$Prognostic,Quantile = 0.5,Ncv=3) ## USING THE FUNCTION Result2 = CVSim(Result, Top = seq(5, 100, by = 5), Data$Survival, Data$Censor,Prognostic = Data$Prognostic) ## GET THE CLASS OF THE OBJECT class(Result2) # An "cvsim" Class## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## GETTING THE cvmm OBJECT Result = CVMetSpecificCoxPh(Fold=3,Survival=Data$Survival, Mdata=t(Data$Mdata),Censor= Data$Censor,Reduce=TRUE,Select=150, Prognostic=Data$Prognostic,Quantile = 0.5,Ncv=3) ## USING THE FUNCTION Result2 = CVSim(Result, Top = seq(5, 100, by = 5), Data$Survival, Data$Censor,Prognostic = Data$Prognostic) ## GET THE CLASS OF THE OBJECT class(Result2) # An "cvsim" Class
Class of object returned by function CVSim.
## S4 method for signature 'cvsim' show(object) ## S4 method for signature 'cvsim' summary(object) ## S4 method for signature 'cvsim,missing' plot(x, y, type = 1, ...)## S4 method for signature 'cvsim' show(object) ## S4 method for signature 'cvsim' summary(object) ## S4 method for signature 'cvsim,missing' plot(x, y, type = 1, ...)
object |
A cvsim class object |
x |
A cvsim class object |
y |
missing |
type |
Plot type. 1 distribution of the HR under test For the Top K metabolites using PCA. 2 distribution of the HR under test For the Top K metabolites using PLS. |
... |
The usual extra arguments to generic functions — see |
HRpcaA 3-way array in which first, second, and third dimensions correspond to number of metabolites, Hazard ratio infromation(Estimated HR, LowerCI and UpperCI), and number of cross validation respectively. This contains the estimated HR on test data and dimension reduction method is PCA.
HRplsA 3-way array in which first, second, and third dimensions correspond to number of metabolites, Hazard ratio infromation(Estimated HR, LowerCI and UpperCI), and number of cross validation respectively. This contains the estimated HR on test data and dimension reduction method is PLS.
NmetsThe number of metabolites in the reduced matrix
NcvThe number of cross validation done
TopA sequence of top k metabolites considered. Default is Top=seq(5,100,by=5)
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
CVPcaPls, SurvPcaClass, SurvPlsClass
## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## FIRST IS THE NETABOLITE BY METABOLITE ANALYSIS w = CVMetSpecificCoxPh(Fold=3,Survival=Data$Survival, Mdata=t(Data$Mdata),Censor= Data$Censor,Reduce=TRUE, Select=150,Prognostic=Data$Prognostic,Quantile = 0.5,Ncv=3) ## USING THE FUNCTION Result = CVSim(w, Top = seq(5, 100, by = 5), Survival=Data$Survival, Censor=Data$Censor, Prognostic = Data$Prognostic) ## GET THE CLASS OF THE OBJECT class(Result) # A "cvsim" Class ## METHOD THAT CAN BE USED FOR THE RESULT show(Result) summary(Result) plot(Result, type =2)## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## FIRST IS THE NETABOLITE BY METABOLITE ANALYSIS w = CVMetSpecificCoxPh(Fold=3,Survival=Data$Survival, Mdata=t(Data$Mdata),Censor= Data$Censor,Reduce=TRUE, Select=150,Prognostic=Data$Prognostic,Quantile = 0.5,Ncv=3) ## USING THE FUNCTION Result = CVSim(w, Top = seq(5, 100, by = 5), Survival=Data$Survival, Censor=Data$Censor, Prognostic = Data$Prognostic) ## GET THE CLASS OF THE OBJECT class(Result) # A "cvsim" Class ## METHOD THAT CAN BE USED FOR THE RESULT show(Result) summary(Result) plot(Result, type =2)
A dataset containing the riskscore, survival parameters (Overall survival and censoring indicator) and other pronostic factors of 149 subjects.
data(DataHR)data(DataHR)
A data frame with 149 rows and 5 variables:
Riskscores of the subjects
Overall survival of the subjects
Censoring indicator for all the patients; 1= Dead and 0 = Alive
The first prognostic factor which is the gender of all the patients; 1=Male and 0 = Female
The second prognostic factor which is the cancer stage of all the patients; 1= Early stage and 0= Advanced stage
...
https://bmccancer.biomedcentral.com/articles/10.1186/s12885-018-4755-1
data(DataHR) summary(DataHR[,1:2])data(DataHR) summary(DataHR[,1:2])
This function generates the null distribution of the HR by permutation approach. Several ways of permutation setting can be implemented. That is, function can be used to generate null distributions for four different validation schemes, PLS based, PCA based, Majority votes based and Lasso based.
DistHR(Survival, Censor, Mdata, Prognostic = NULL, Quantile = 0.5, Reduce = FALSE, Select = 15, nperm = 100, case = 2, Validation = c("PLSbased", "PCAbased", "L1based", "MVbased"))DistHR(Survival, Censor, Mdata, Prognostic = NULL, Quantile = 0.5, Reduce = FALSE, Select = 15, nperm = 100, case = 2, Validation = c("PLSbased", "PCAbased", "L1based", "MVbased"))
Survival |
A vector of survival time with length equals to number of subjects |
Censor |
A vector of censoring indicator |
Mdata |
A large or small metabolic profile matrix. A matrix with metabolic profiles where the number of rows should be equal to the number of metabolites and number of columns should be equal to number of patients. |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
Quantile |
The cut off value for the classifier, default is the median cutoff |
Reduce |
A boolean parameter indicating if the metabolic profile matrix should be reduced, default is TRUE and larger metabolic profile matrix is reduced by supervised pca approach and first pca is extracted from the reduced matrix to be used in the classifier. |
Select |
Number of metabolites (default is 15) to be selected from supervised PCA. This is valid only if the argument Reduce=TRUE |
nperm |
Number of permutations to be used and default 100 |
case |
There are seven different ways on how to call this argument:
|
Validation |
There are four different validation schemes where the null distribution can be estimated. That is c("PLSbased","PCAbased","L1based","MVbased"). |
This function generates the null distribution of the HR by permutation approach either using a large metabolite matrix or a reduced version by supervised pca approach. Several ways of permutation setting can be implemented. That is, the function can be used to generate null distributions for four different validation schemes which are PLS based, PCA based, Majority votes based and Lasso based. Note this function internally calls function SurvPcaClass, SurvPlsClass, Majorityvotes, and Lasoelacox.
A object of class perm is returned with the following values
HRobs |
Estimated HR for low risk group on the original data |
HRperm |
Estimated HR for low risk group on the permuted data |
nperm |
Number of permutations carried out |
Validation |
The validation scheme that was used |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
coxph, EstimateHR, SurvPcaClass, SurvPlsClass, Majorityvotes, Lasoelacox, EstimateHR, Lasoelacox
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Example <- DistHR(Survival = Data$Survival,Mdata = t(Data$Mdata), Censor = Data$Censor,Reduce=FALSE,Select=15,Prognostic=Data$Prognostic, Quantile = 0.5, nperm=10, case=2, Validation=c("L1based"))## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Example <- DistHR(Survival = Data$Survival,Mdata = t(Data$Mdata), Censor = Data$Censor,Reduce=FALSE,Select=15,Prognostic=Data$Prognostic, Quantile = 0.5, nperm=10, case=2, Validation=c("L1based"))
The Function classifies subjects into Low and High risk group using the risk scores based on the cut-off percentile.It also visualize survival fit along with HR estimates.
EstimateHR(Risk.Scores, Data.Survival, Prognostic = NULL, Plots = FALSE, Quantile = 0.5)EstimateHR(Risk.Scores, Data.Survival, Prognostic = NULL, Plots = FALSE, Quantile = 0.5)
Risk.Scores |
A vector of risk scores with size equals to number of subjects |
Data.Survival |
A dataframe in which the first column is the Survival and the second column is the Censoring indicator for each subject. |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect |
Plots |
A boolean parameter indicating if plots should be shown. Default is FALSE |
Quantile |
The cut off value for the classifier, default is the median cutoff |
The risk scores obtained using the signature is then used to generate the risk group by dividing subjects into low and high risk group. A Cox model is then fitted with the risk group as covariate in the presence or absence of prognostic factors and or treatment effect. The extent of survival in the risk groups is known
An object of is returned, which is a list with the results of the cox regression and some informative plot concerning survival of the risk group.
SurvResult |
The cox proportional regression result |
Riskgroup |
The riskgroup based on the riskscore and the cut off value and length is equal to number of subjects |
KMplot |
The Kaplan-Meier survival plot of the riskgroup |
SurvBPlot |
The distribution of the survival in the riskgroup |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
### Classification and estimating with prognostic factors data(DataHR) Result = EstimateHR(Risk.Scores=DataHR[,1],Data.Survival=DataHR[,2:3] ,Prognostic=DataHR[,4:5],Plots=FALSE,Quantile=0.50) ### Classification and estimating without prognostic factors data(DataHR) Result = EstimateHR(Risk.Scores=DataHR[,1],Data.Survival=DataHR[,2:3] ,Prognostic=NULL,Plots=FALSE,Quantile=0.50)### Classification and estimating with prognostic factors data(DataHR) Result = EstimateHR(Risk.Scores=DataHR[,1],Data.Survival=DataHR[,2:3] ,Prognostic=DataHR[,4:5],Plots=FALSE,Quantile=0.50) ### Classification and estimating without prognostic factors data(DataHR) Result = EstimateHR(Risk.Scores=DataHR[,1],Data.Survival=DataHR[,2:3] ,Prognostic=NULL,Plots=FALSE,Quantile=0.50)
Class of object returned by function Icvlasoel.
## S4 method for signature 'fcv' show(object) ## S4 method for signature 'fcv' summary(object) ## S4 method for signature 'fcv,missing' plot(x, y, type = 1, ...)## S4 method for signature 'fcv' show(object) ## S4 method for signature 'fcv' summary(object) ## S4 method for signature 'fcv,missing' plot(x, y, type = 1, ...)
object |
A fcv class object |
x |
A fcv class object |
y |
missing |
type |
Plot type. 1 is the distribution of the inner cross validated HR under test data for each outer iterations and estimated HR on the out of bag data are superimposed. 2 Estimated HR Density for low Risk Group . |
... |
The usual extra arguments to generic functions — see |
RuntimeA vector of runtime for each iteration measured in seconds.
FoldNumber of folds used.
NcvNumber of outer cross validations used.
NicvNumber of inner cross validations used.
TopKThe Top metabolites used
HRInnerA 3-way array in which first, second, and third dimensions correspond to Nicv, 1, and Ncv respectively. This contains estimated HR for low risk group on the out of bag data.
HRTestA matrix of survival information for the test dataset based on the out of bag data. It has three columns representing the estimated HR, the 95% lower confidence interval and the 95% upper confidence interval.
WeightA matrix with columns equals number of TopK metabolites and rows Ncv. Note that Weights are estimated as colMeans of coefficients matrix return from the inner cross validations.
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
CVLasoelacox,
EstimateHR, glmnet, Lasoelacox
## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## USE THE FUNCTION Eg = Icvlasoel(Data$Survival, Data$Censor, Data$Prognostic, t(Data$Mdata), Fold = 3,Ncv = 5, Nicv = 7, Alpha = 1, TopK = colnames(Data$Mdata[,80:100]), Weights = FALSE) ## GET THE CLASS OF THE OBJECT class(Eg) # An "fcv" Class ## METHOD THAT CAN BE USED FOR THIS CLASS show(Eg) summary(Eg) plot(Eg, type =1)## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## USE THE FUNCTION Eg = Icvlasoel(Data$Survival, Data$Censor, Data$Prognostic, t(Data$Mdata), Fold = 3,Ncv = 5, Nicv = 7, Alpha = 1, TopK = colnames(Data$Mdata[,80:100]), Weights = FALSE) ## GET THE CLASS OF THE OBJECT class(Eg) # An "fcv" Class ## METHOD THAT CAN BE USED FOR THIS CLASS show(Eg) summary(Eg) plot(Eg, type =1)
The function does cross validation for Lasso, Elastic net and Ridge regressions models based on fixed or top selected metabolites from CVLasoelacox with classifier validated on a independent sample for the survial analysis and classification. The survival analysis is based on the selected metabolites in the presence or absene of prognostic factors.
Icvlasoel(Survival, Censor, Prognostic = NULL, Mdata, Fold = 3, Ncv = 50, Nicv = 100, Alpha = 0.1, TopK, Weights = FALSE)Icvlasoel(Survival, Censor, Prognostic = NULL, Mdata, Fold = 3, Ncv = 50, Nicv = 100, Alpha = 0.1, TopK, Weights = FALSE)
Survival |
A vector of survival time with length equals to number of subjects |
Censor |
A vector of censoring indicator |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
Mdata |
A large or small metabolic profile matrix. A matrix with metabolic profiles where the number of rows should be equal to the number of metabolites and number of columns should be equal to number of patients. |
Fold |
number of folds to be used for the cross validation. Its value ranges between 3 and the numbe rof subjects in the dataset |
Ncv |
Number of validations to be carried out. The default is 25. |
Nicv |
Number of validations to be carried out for the inner loop. The default is 5. |
Alpha |
The mixing parameter for glmnet (see |
TopK |
Top list of metabolites. Usually this can be mostly selected metabolites by function |
Weights |
A logical flag indicating if a fixed or non-fixed weights should be used during the classifier evaluations. Default is FALSE. |
The function does cross validation for Lasso, Elastic net and Ridge regressions models based on fixed or top selected metabolites from CVLasoelacox with classifier validated on a independent sample for the survial analysis and classification. The survival analysis is based on the selected metabolites in the presence or absene of prognostic factors. The classifier is built on the weights obtain from the inner cross validations results and it is tested on out-of-bag data. These weights can be fixed or can be updated at each outer iteration. If weights are not fixed then patients are classified using majority votes. Otherwise, weights obtained from the inner cross validations are summarized by mean weights and used in the classifier. Inner cross validations are performed by calling to function CVLasoelacox. Hazard ratio for low risk group is estimated using out-of-bag data.
A object of class fcv is returned with the following values
Runtime |
A vector of runtime for each iteration measured in seconds. |
Fold |
Number of folds used. |
Ncv |
Number of outer cross validations used. |
Nicv |
Number of inner cross validations used. |
TopK |
The Top metabolites used |
HRInner |
A 3-way array in which first, second, and third dimensions correspond to Nicv, 1, and Ncv respectively. This contains estimated HR for low risk group on the out of bag data. |
HRTest |
A matrix of survival information for the test dataset based on the out of bag data. It has three columns representing the estimated HR, the 95% lower confidence interval and the 95% upper confidence interval. |
Weight |
A matrix with columns equals number of TopK metabolites and rows Ncv. Note that Weights are estimated as colMeans of coefficients matrix return from the inner cross validations. |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
CVLasoelacox,
EstimateHR, glmnet, Lasoelacox
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Results = Icvlasoel(Data$Survival, Data$Censor, Data$Prognostic, t(Data$Mdata), Fold = 3,Ncv = 5, Nicv = 7, Alpha = 1, TopK = colnames(Data$Mdata[,80:100]), Weights = FALSE) ## NUMBER OF Outer CV Results@Ncv ## NUMBER OF Inner CV Results@Nicv ## HR of low risk group for the Inner CV Results@HRInner ## HR of low risk group for the out of bag dataset Results@HRTest ## The weight for the analysis Results@Weight## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Results = Icvlasoel(Data$Survival, Data$Censor, Data$Prognostic, t(Data$Mdata), Fold = 3,Ncv = 5, Nicv = 7, Alpha = 1, TopK = colnames(Data$Mdata[,80:100]), Weights = FALSE) ## NUMBER OF Outer CV Results@Ncv ## NUMBER OF Inner CV Results@Nicv ## HR of low risk group for the Inner CV Results@HRInner ## HR of low risk group for the out of bag dataset Results@HRTest ## The weight for the analysis Results@Weight
The function uses the glmnet function to firstly do the variable selection either with Lasso, Elastic net or ridge regressions before the survial analysis. The survival analysis is based on the selected metabolites in the presence or absene of prognostic factors.
Lasoelacox(Survival, Censor, Mdata, Prognostic, Quantile = 0.5, Metlist = NULL, Plots = FALSE, Standardize = TRUE, Alpha = 1, Fold = 4, nlambda = 100)Lasoelacox(Survival, Censor, Mdata, Prognostic, Quantile = 0.5, Metlist = NULL, Plots = FALSE, Standardize = TRUE, Alpha = 1, Fold = 4, nlambda = 100)
Survival |
A vector of survival time with length equals to number of subjects |
Censor |
A vector of censoring indicator |
Mdata |
A large or small metabolic profile matrix. A matrix with metabolic profiles where the number of rows should be equal to the number of metabolites and number of columns should be equal to number of patients. |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
Quantile |
The cut off value for the classifier, default is the median cutoff |
Metlist |
A list of metabolites to be considered in the model usually smaller than the metabolites in the Mdata . Default is to use all metabolites available |
Plots |
A boolean parameter indicating if plots should be shown. Default is FALSE. If TRUE, the first plot is the partial likelihood deviance against the logarithmn of each lambda while the second is the coefficients versus the lamdas |
Standardize |
A Logical flag for the standardization of the metabolite matrix, prior to fitting the model sequence. The coefficients are always returned on the original scale. Default is standardize=TRUE. |
Alpha |
The mixing parameter for glmnet (see |
Fold |
number of folds to be used for the cross validation. Its value ranges between 3 and the numbe rof subjects in the dataset |
nlambda |
The number of lambda values - default is 100 as in glmnet. |
This is a wrapper function for glmnet and it fits models using either Lasso, Elastic net and Ridge regressions. This is done in the presence or absene of prognostic factors. The prognostic factor when avaialable will always be forced to be in the model so no penalty for it. Optimum lambda will be used to select the non-zero shrinkage coefficients, the nonzero selceted metabolites will thus be used in the survival analysis and in calculation of the risk scores.
A object is returned with the following values
Coefficients.NonZero |
The coefficients of the selected metabolites |
Selected.Mets |
The selected metabolites |
n |
The number of selected metabolites |
Risk.scores |
The risk scores of the subjects |
Risk.group |
The risk classification of the subjects based on the specified quantile |
SurvFit |
The cox analysis of the riskgroup based on the selected metabolites and the prognostic factors |
Select |
A Boolean argument indicating if there was selection or not |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Results = Lasoelacox(Survival=Data$Survival, Censor=Data$Censor, Mdata=t(Data$Mdata), Prognostic = Data$Prognostic, Quantile = 0.5, Metlist = NULL, Plots = FALSE, Standardize = TRUE, Alpha = 1) ## VIEW THE SELECTED METABOLITES Results$Selected.mets ## NUMBER OF SELECTED METABOLITES Results$n ## VIEW THE CLASSIFICATION GROUP OF EACH SUBJECT Results$Risk.Group ## VIEW THE SURVIVAL ANALYSIS RESULT Results$SurvFit ## TO CHECK IF THERE WAS ANY SELECTION Results$Select## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Results = Lasoelacox(Survival=Data$Survival, Censor=Data$Censor, Mdata=t(Data$Mdata), Prognostic = Data$Prognostic, Quantile = 0.5, Metlist = NULL, Plots = FALSE, Standardize = TRUE, Alpha = 1) ## VIEW THE SELECTED METABOLITES Results$Selected.mets ## NUMBER OF SELECTED METABOLITES Results$n ## VIEW THE CLASSIFICATION GROUP OF EACH SUBJECT Results$Risk.Group ## VIEW THE SURVIVAL ANALYSIS RESULT Results$SurvFit ## TO CHECK IF THERE WAS ANY SELECTION Results$Select
The Function fits cox proportional hazard model and does classification based on the majority votes.
Majorityvotes(Result, Prognostic, Survival, Censor, J = 1)Majorityvotes(Result, Prognostic, Survival, Censor, J = 1)
Result |
An object obtained from the metabolite specific analysis ( |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
Survival |
A vector of survival time with length equals to number of subjects |
Censor |
A vector of censoring indicator |
J |
The jth set of patients required for the visualization. The default is J=1 which is the first set of patients. For visualization, J should be less than the number of patients divided by 25 |
The Function fits cox proportional hazard model and does classification based on the majority votes while estimating the Hazard ratio of the low risk group. The function firstly count the number of low risk classification for each subject based on the metabolite specific analysis which determines the majority votes. In addition, It visualizes the metabolic specific calssification for the subjects. 25 subjects is taken for visualization purpose.
A list is returned with the following values
Model.result |
The cox proportional regression result based on the majority vote classification |
N |
The majority vote for each subject |
Classif |
The majority vote classification for each subjects |
Group |
The classification of the subjects based on each metabolite analysis |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
Hastie T, Tibshirani R, Friedman J (2001). The elements of statistical learning: data mining, inference, and prediction: with 200 full-color illustrations. New York: Springer-Verlag.
MSpecificCoxPh, coxph, EstimateHR
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## RUNNING THE METABOLITE SPECIFIC FUNCTION Example1 = MSpecificCoxPh(Survival = Data$Survival, Mdata = t(Data$Mdata), Censor = Data$Censor, Reduce = FALSE, Select = 15,Prognostic = Data$Prognostic, Quantile = 0.5) ## USING THE FUNCTION Result2 = Majorityvotes(Example1,Data$Prognostic, Data$Survival,Data$Censor,J=2) ## THE SURVIVAL ANALYSIS FOR MAJORITY VOTE RESULT Result2$Model.result ### THE MAJORITY VOTE FOR EACH SUBJECT Result2$N ### THE MAJORITY VOTE CLASSIFICATION FOR EACH SUBJECT Result2$Classif ### THE GROUP FOR EACH SUBJECT BASED ON THE METABOLITE SPPECIFIC ANALYSIS Result2$Group## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## RUNNING THE METABOLITE SPECIFIC FUNCTION Example1 = MSpecificCoxPh(Survival = Data$Survival, Mdata = t(Data$Mdata), Censor = Data$Censor, Reduce = FALSE, Select = 15,Prognostic = Data$Prognostic, Quantile = 0.5) ## USING THE FUNCTION Result2 = Majorityvotes(Example1,Data$Prognostic, Data$Survival,Data$Censor,J=2) ## THE SURVIVAL ANALYSIS FOR MAJORITY VOTE RESULT Result2$Model.result ### THE MAJORITY VOTE FOR EACH SUBJECT Result2$N ### THE MAJORITY VOTE CLASSIFICATION FOR EACH SUBJECT Result2$Classif ### THE GROUP FOR EACH SUBJECT BASED ON THE METABOLITE SPPECIFIC ANALYSIS Result2$Group
This package develope biomarker signature for metabolic data. It contains a set of functions and cross validation methods to validate and select biomarkers when the outcome of interest is survival. The package can handle prognostic factors and mainly metabolite matrix as input, the package can served as biomarker validation tool.
It can be used with any form of high dimensional/omics data such as: Metabolic data, Gene expression matrix, incase you dont have a data it can simulate hypothetical scinerio of a high dimensional data based on the desired biological parameters
It developed any form of signature from the high dimensional data to be used for other purpose
It also employs data reduction techniques such as PCA, PLS and Lasso
It classifies subjects based on the signatures into Low and high risk group
It incorporate the use of subject prognostic information for the to enhance the biomarker for classification
It gives information about the surival rate of subjects depending on the classification
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
The function selects the frquency of selection from the shrinkage method (LASSO, Elastic-net) based on cross validation, that is the number of times each metabolite occur during the cross-validation process. In case of large metabolomic matrix then the N argument can be used to select metabolites occurence at a particular frequency.
MetFreq(Object, TopK = 20, N = 3)MetFreq(Object, TopK = 20, N = 3)
Object |
An object of class |
TopK |
The number of Top K metabolites (5 by default) to be displayed in the frequency of selection graph. |
N |
The metqbolites with the specified frequency should be displayed in the frequency of selection graph. |
This function outputs the mostly selected metabolites during the LASSO and Elastic-net cross validation. Selected top metabolites are ranked based on frequency of selection and also a particular frequency cqn be selected. In addition, it visualizes the selected top metabolites based on the minimum frequency specified.
A vector of metabolites and their frequency of selection. Also, a graphical representation is displayed.
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
cvmm, coxph,
EstimateHR,CVLasoelacox
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## CROSS-VALIDATION FOR LASSO AND ELASTIC-NET Result = CVLasoelacox(Survival = Data$Survival, Censor = Data$Censor, Mdata = t(Data$Mdata), Prognostic = Data$Prognostic, Quantile = 0.5, Metlist = NULL,Standardize = TRUE, Reduce=FALSE, Select=15, Alpha = 1,Fold = 4,Ncv = 10,nlambda = 100) ## CONFIRMING THE CLASS class(Result) ## USING THE FUNCTION MetFreq(Result,TopK = 5, N=5)## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## CROSS-VALIDATION FOR LASSO AND ELASTIC-NET Result = CVLasoelacox(Survival = Data$Survival, Censor = Data$Censor, Mdata = t(Data$Mdata), Prognostic = Data$Prognostic, Quantile = 0.5, Metlist = NULL,Standardize = TRUE, Reduce=FALSE, Select=15, Alpha = 1,Fold = 4,Ncv = 10,nlambda = 100) ## CONFIRMING THE CLASS class(Result) ## USING THE FUNCTION MetFreq(Result,TopK = 5, N=5)
Class of object returned by function MSpecificCoxPh.
## S4 method for signature 'ms' show(object) ## S4 method for signature 'ms' summary(object) ## S4 method for signature 'ms,ANY' plot(x, y, ...)## S4 method for signature 'ms' show(object) ## S4 method for signature 'ms' summary(object) ## S4 method for signature 'ms,ANY' plot(x, y, ...)
object |
A ms class object |
x |
A ms class object |
y |
missing |
... |
The usual extra arguments to generic functions — see |
plot signature(x = "ms"): Plots for ms class analysis results signature(x = "ms"): Plots for ms class analysis results.
Any parameters of plot.default may be passed on to this particular plot method.
show(ms-object)
ResultA list of dataframes of each output object of coxph for the metabolites.
HRRGA dataframe with estimated metabolite-specific HR for low risk group and 95 percent CI.
GroupA matrix of the classification group a subject belongs to for each of the metabolite analysis. The metabolites are on the rows and the subjects are the columns
MetnamesThe names of the metabolites for the analysis
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## DO THE METABOLITE BY METABOLITE ANALYSIS Eg = MSpecificCoxPh(Survival=Data$Survival, Mdata=t(Data$Mdata), Censor=Data$Censor, Reduce = FALSE, Select = 15, Prognostic=Data$Prognostic, Quantile = 0.5) ## GET THE CLASS OF THE OBJECT class(Eg) # An "ms" Class ## METHOD THAT CAN BE USED FOR THIS CLASS show(Eg) summary(Eg) plot(Eg)## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## DO THE METABOLITE BY METABOLITE ANALYSIS Eg = MSpecificCoxPh(Survival=Data$Survival, Mdata=t(Data$Mdata), Censor=Data$Censor, Reduce = FALSE, Select = 15, Prognostic=Data$Prognostic, Quantile = 0.5) ## GET THE CLASS OF THE OBJECT class(Eg) # An "ms" Class ## METHOD THAT CAN BE USED FOR THIS CLASS show(Eg) summary(Eg) plot(Eg)
The Function generates metabolic profile of any number of patients and also their survival information.
MSData(nPatients = 100, nMet = 150, Prop = 0.5)MSData(nPatients = 100, nMet = 150, Prop = 0.5)
nPatients |
The number of patients |
nMet |
The number of metabolites |
Prop |
The proportion of patients having low risk |
The function generates the metabolic profile where small set of metabolites (30) are informative and rest of them are set as noisy metabolites. Next to that Survival time and Censoring information are generated based on first right singular vectors of svd of the metabolic profile matrix. It also generates other prognostic factors such as Age, Stage and Gender which are slightly correlated with survival time.
An object of class list is returned with the following items .
Censor |
The censoring/event indicator |
Survival |
The Survival time |
Met.names |
The vector of metabolites |
Mdata |
The metabolic profile matrix |
Prognostic |
A data frame with prognostic factors. |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
#GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) SurvTime<-Data$Survival Censor<-Data$Censor ProgFact<-Data$Prognostic MetData<-Data$Mdata Metnames<-Data$Met.names#GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) SurvTime<-Data$Survival Censor<-Data$Censor ProgFact<-Data$Prognostic MetData<-Data$Mdata Metnames<-Data$Met.names
The Function fits cox proportional hazard model and does classification for each metabolite
MSpecificCoxPh(Survival, Mdata, Censor, Reduce = FALSE, Select = 15, Prognostic = NULL, Quantile = 0.5)MSpecificCoxPh(Survival, Mdata, Censor, Reduce = FALSE, Select = 15, Prognostic = NULL, Quantile = 0.5)
Survival |
A vector of survival time with length equals to number of subjects |
Mdata |
A large or small metabolic profile matrix. A matrix with metabolic profiles where the number of rows should be equal to the number of metabolites and number of columns should be equal to number of patients. |
Censor |
A vector of censoring indicator |
Reduce |
A boolean parameter indicating if the metabolic profile matrix should be reduced, default is TRUE and larger metabolic profile matrix is reduced by supervised pca approach and first pca is extracted from the reduced matrix to be used in the classifier. |
Select |
Number of metabolites (default is 15) to be selected from supervised PCA. This is valid only if the argument Reduce=TRUE |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
Quantile |
The cut off value for the classifier, default is the median cutoff |
This function fits metabolite by metabolite Cox proportional hazard model and perform the classification based on a speciied quantile risk score which has been estimated using a single metabolite. Function is useful for majority vote classification method and metabolite by metabolite analysis and also for top K metabolites.
A object of class ms is returned with the following values
Result |
The cox proportional regression result for each metabolite |
HRRG |
The hazard ratio statistics (Hazard-ratio, Lower confidence interval and upper confidence interval) of the riskgroup based on the riskscore and the cut off value for each metabolite |
Group |
The classification of the subjects based on each metabolite analysis |
Metnames |
The names of the metabolites for the analysis |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Example1 = MSpecificCoxPh(Survival = Data$Survival, Mdata = t(Data$Mdata), Censor = Data$Censor, Reduce = FALSE, Select = 15,Prognostic = Data$Prognostic, Quantile = 0.5) ## KNOWLING THE CLASS OF THE OUTPUT class(Example1) ## EXTRACTING THE COMPONENT OF THE FUNCTION ### HAZARD RATIO INFORMATION FOR EACH METABOLITES Example1@HRRG ### COX MODEL RESULT FOR EACH METABOLITES Example1@Result ### CLASSIFICATION FOR EACH METABOLITES Example1@Group## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Example1 = MSpecificCoxPh(Survival = Data$Survival, Mdata = t(Data$Mdata), Censor = Data$Censor, Reduce = FALSE, Select = 15,Prognostic = Data$Prognostic, Quantile = 0.5) ## KNOWLING THE CLASS OF THE OUTPUT class(Example1) ## EXTRACTING THE COMPONENT OF THE FUNCTION ### HAZARD RATIO INFORMATION FOR EACH METABOLITES Example1@HRRG ### COX MODEL RESULT FOR EACH METABOLITES Example1@Result ### CLASSIFICATION FOR EACH METABOLITES Example1@Group
Class of object returned by function DistHR.
## S4 method for signature 'perm' show(object) ## S4 method for signature 'perm' summary(object) ## S4 method for signature 'perm,ANY' plot(x, y, ...)## S4 method for signature 'perm' show(object) ## S4 method for signature 'perm' summary(object) ## S4 method for signature 'perm,ANY' plot(x, y, ...)
object |
A perm class object |
x |
A perm class object |
y |
missing |
... |
The usual extra arguments to generic functions — see |
HRobsEstimated HR for low risk group on the original data.
HRpermEstimated HR for low risk group on the permuted data
npermNumber of permutations carried out.
ValidationThe validation scheme that was used.
The first, third and last vertical line on the plot are the lower, median and upper CI of the permuted data estimated HR while the red line is the estimated HR of the original data
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
DistHR, EstimateHR, SurvPcaClass, SurvPlsClass, Majorityvotes, Lasoelacox, EstimateHR, Lasoelacox
## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## USING THE FUNCTION Example <- DistHR(Survival = Data$Survival,Mdata = t(Data$Mdata), Censor = Data$Censor,Reduce=FALSE,Select=15,Prognostic=Data$Prognostic, Quantile = 0.5, nperm=10, case=2, Validation=c("L1based")) ## GET THE CLASS OF THE OBJECT class(Example) # A "perm" Class ## METHOD THAT CAN BE USED FOR THIS CLASS show(Example) summary(Example) plot(Example)## GENERATE SOME METABOLIC SURVIVAL DATA WITH PROGNOSTIC FACTORS Data<-MSData(nPatients=100,nMet=150,Prop=0.5) ## USING THE FUNCTION Example <- DistHR(Survival = Data$Survival,Mdata = t(Data$Mdata), Censor = Data$Censor,Reduce=FALSE,Select=15,Prognostic=Data$Prognostic, Quantile = 0.5, nperm=10, case=2, Validation=c("L1based")) ## GET THE CLASS OF THE OBJECT class(Example) # A "perm" Class ## METHOD THAT CAN BE USED FOR THIS CLASS show(Example) summary(Example) plot(Example)
The function performs sensitivity of the cut off quantile for obtaining the risk group obtained under SurvPlsClass, SurvPcaClass or Lasoelacox requires for the survival analysis and classification.
QuantileAnalysis(Survival, Mdata, Censor, Reduce = TRUE, Select = 150, Prognostic = NULL, Plots = FALSE, DimMethod = c("PLS", "PCA", "SM"), Alpha = 1)QuantileAnalysis(Survival, Mdata, Censor, Reduce = TRUE, Select = 150, Prognostic = NULL, Plots = FALSE, DimMethod = c("PLS", "PCA", "SM"), Alpha = 1)
Survival |
A vector of survival time with length equals to number of subjects |
Mdata |
A large or small metabolic profile matrix. A matrix with metabolic profiles where the number of rows should be equal to the number of metabolites and number of columns should be equal to number of patients. |
Censor |
A vector of censoring indicator |
Reduce |
A boolean parameter indicating if the metabolic profile matrix should be reduced, default is TRUE and larger metabolic profile matrix is reduced by supervised pca approach and first pca is extracted from the reduced matrix to be used in the classifier. |
Select |
Number of metabolites (default is 15) to be selected from supervised PCA. This is valid only if th argument Reduce=TRUE |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
Plots |
A boolean parameter indicating if the graphical represenataion of the analysis should be shown. Default is FALSE and it is only valid for the PCA or PLS dimension method. |
DimMethod |
The dimension method to be used. PCA implies using the |
Alpha |
The mixing parameter for glmnet (see |
This function investigates how each analysis differs from the general median cutoff of 0.5, therefore to see the sensitive nature of the survival result different quantiles ranging from 10th percentile to 90th percentiles were used. The sensitive nature of the quantile is investigated under SurvPlsClass, SurvPcaClass or Lasoelacox while relate to the 3 different Dimension method to select from.
A Dataframe is returned depending on weather a data reduction method should be used or not. The dataframe contains the HR of the low risk group for each percentile.
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
Bair E, Hastie T, Debashis P, Tibshirani R (2006). “Prediction by supervised principal components.” American Statistics Association,, 101(473), 119-137.
coxph,EstimateHR,
SurvPcaClass,
SurvPlsClass,Lasoelacox
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE PCA METHOD Result = QuantileAnalysis(Data$Survival,t(Data$Mdata), Data$Censor,Reduce=FALSE, Select=150, Prognostic=Data$Prognostic, Plots = TRUE,DimMethod="PCA",Alpha=1) ## USING THE PLS METHOD Result = QuantileAnalysis(Data$Survival,t(Data$Mdata), Data$Censor,Reduce=FALSE, Select=150, Prognostic=Data$Prognostic, Plots = TRUE,DimMethod="PLS",Alpha=1) ## USING THE SM METHOD Result = QuantileAnalysis(Data$Survival,t(Data$Mdata), Data$Censor,Reduce=FALSE, Select=150, Prognostic=Data$Prognostic, Plots = FALSE,DimMethod="SM",Alpha=1)## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE PCA METHOD Result = QuantileAnalysis(Data$Survival,t(Data$Mdata), Data$Censor,Reduce=FALSE, Select=150, Prognostic=Data$Prognostic, Plots = TRUE,DimMethod="PCA",Alpha=1) ## USING THE PLS METHOD Result = QuantileAnalysis(Data$Survival,t(Data$Mdata), Data$Censor,Reduce=FALSE, Select=150, Prognostic=Data$Prognostic, Plots = TRUE,DimMethod="PLS",Alpha=1) ## USING THE SM METHOD Result = QuantileAnalysis(Data$Survival,t(Data$Mdata), Data$Censor,Reduce=FALSE, Select=150, Prognostic=Data$Prognostic, Plots = FALSE,DimMethod="SM",Alpha=1)
The Function fits cox proportional hazard model and does classification by sequentially increasing the metabolites using either PCA or PLS based on the topK metabolites specified.
SIMet(TopK = 15, Survival, Mdata, Censor, Reduce = TRUE, Select = 50, Prognostic = NULL, Plot = FALSE, DimMethod = c("PLS", "PCA"), ...)SIMet(TopK = 15, Survival, Mdata, Censor, Reduce = TRUE, Select = 50, Prognostic = NULL, Plot = FALSE, DimMethod = c("PLS", "PCA"), ...)
TopK |
Top K metabolites (15 by default) to be used in the sequential analysis. |
Survival |
A vector of survival time with length equals to number of subjects |
Mdata |
A large or small metabolic profile matrix. A matrix with metabolic profiles where the number of rows should be equal to the number of metabolites and number of columns should be equal to number of patients. |
Censor |
A vector of censoring indicator |
Reduce |
A boolean parameter indicating if the metabolic profile matrix should be reduced, default is TRUE and larger metabolic profile matrix is reduced by supervised pca approach and first pca is extracted from the reduced matrix to be used in the classifier. |
Select |
Number of metabolites to be selected from supervised PCA. This is valid only if the argument Reduce=TRUE |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
Plot |
A boolean parameter indicating if Plot should be shown. Default is FALSE |
DimMethod |
Dimension reduction method which can either be PLS or PCA. |
... |
Additinal arguments for plotting and only valid if Plot=TRUE |
This function sequentially increase the number of top K metabolites to be used in the PCA or PLS methods in order to obtain the risk score. This function internally calls MSpecificCoxPh to rank the metabolites based on HR for each metabolite. Therefore metabolites can be ordered based on increasing order of the HR for low risk group. Thereafter, the function takes few top K (15 is the default) to be used in the sequential analysis.
A list containing a data frame with estimated HR along with 95% CI at each TopK value for the sequential analysis.
Result |
The hazard ratio statistics (HR, Lower confidence interval and upper confidence interval) of the lower riskgroup based for each sequential metabolite analysis |
TopKplot |
A graphical representation of the Result containing the hazard ratio statistics |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
Vinzi VE, Chin WW, Henseler J, Wang H (2010). Handbook of Partial Least Squares: Concepts, Methods and Applications, 1st edition. Springer Publishing Company, Incorporated.
Bair E, Hastie T, Debashis P, Tibshirani R (2006). “Prediction by supervised principal components.” American Statistics Association,, 101(473), 119-137.
coxph, EstimateHR, MSpecificCoxPh, SurvPcaClass, SurvPcaClass
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Example1 = SIMet(TopK = 10, Survival=Data$Survival, Mdata=t(Data$Mdata), Censor=Data$Censor, Reduce = TRUE, Select = 50,Prognostic = Data$Prognostic, Plot = TRUE, DimMethod ="PLS") ## FOR THE HR STATISTICS Example1$Result ## FOR THE GRAPHICAL OUTPUT Example1$TopKplot## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Example1 = SIMet(TopK = 10, Survival=Data$Survival, Mdata=t(Data$Mdata), Censor=Data$Censor, Reduce = TRUE, Select = 50,Prognostic = Data$Prognostic, Plot = TRUE, DimMethod ="PLS") ## FOR THE HR STATISTICS Example1$Result ## FOR THE GRAPHICAL OUTPUT Example1$TopKplot
The function performs principal component analysis (PCA) on Metabolomics matrix and fit Cox proportional hazard model with covariates using also the first PCA as covariates.
SurvPcaClass(Survival, Mdata, Censor, Reduce = TRUE, Select = 150, Prognostic = NULL, Plots = FALSE, Quantile = 0.5)SurvPcaClass(Survival, Mdata, Censor, Reduce = TRUE, Select = 150, Prognostic = NULL, Plots = FALSE, Quantile = 0.5)
Survival |
A vector of survival time with length equals to number of subjects |
Mdata |
A large or small metabolic profile matrix. A matrix with metabolic profiles where the number of rows should be equal to the number of metabolites and number of columns should be equal to number of patients. |
Censor |
A vector of censoring indicator |
Reduce |
A boolean parameter indicating if the metabolic profile matrix should be reduced, default is TRUE and larger metabolic profile matrix is reduced by supervised pca approach and first pca is extracted from the reduced matrix to be used in the classifier. |
Select |
Number of metabolites (default is 15) to be selected from supervised PCA. This is valid only if th argument Reduce=TRUE |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
Plots |
A boolean parameter indicating if the plots should be shown. Default is FALSE |
Quantile |
The cut off value for the classifier, default is the median cutoff |
This function can also be used to perform the grid analysis where the grid will be several quantile values and default is 0.5 which is the median cut-off. This function can handle single and multiple metabolites. For larger Metabolomics matrix, this function will reduce largerMetabolomics matrix to smaller version using supervised pca approach and this is by default done and can be control by using the argument Reduce. Other prognostic factors can be included to the model.
A object of class SurvPca is returned with the following values
Survfit |
The cox proportional regression result using the first PCA |
Riskscores |
A vector of risk scores which is equal to the number of patents. |
Riskgroup |
The classification of the subjects based on the PCA into low or high risk group |
pc1 |
The First PCA scores based on either the reduced Metabolite matrix or the full matrix |
KMplot |
The Kaplan-Meier survival plot of the riskgroup |
SurvBPlot |
The distribution of the survival in the riskgroup |
Riskpca |
The plot of Risk scores vs first PCA |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
Bair E, Hastie T, Debashis P, Tibshirani R (2006). “Prediction by supervised principal components.” American Statistics Association,, 101(473), 119-137.
coxph,
EstimateHR, princomp,
SurvPlsClass
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Result = SurvPcaClass(Survival=Data$Survival, Mdata=t(Data$Mdata), Censor=Data$Censor, Reduce = FALSE, Select = 150, Prognostic = Data$Prognostic, Plots = FALSE, Quantile = 0.5) ## GETTING THE SURVIVAL REGRESSION OUTPUT Result$SurvFit ## GETTING THE RISKSCORES Result$Riskscores ### GETTING THE RISKGROUP Result$Riskgroup ### OBTAINING THE FIRST PRINCIPAL COMPONENT SCORES Result$pc1## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Result = SurvPcaClass(Survival=Data$Survival, Mdata=t(Data$Mdata), Censor=Data$Censor, Reduce = FALSE, Select = 150, Prognostic = Data$Prognostic, Plots = FALSE, Quantile = 0.5) ## GETTING THE SURVIVAL REGRESSION OUTPUT Result$SurvFit ## GETTING THE RISKSCORES Result$Riskscores ### GETTING THE RISKGROUP Result$Riskgroup ### OBTAINING THE FIRST PRINCIPAL COMPONENT SCORES Result$pc1
The function performs partial least squares (PLS) and principal component regression on Metabolomics matrix and fit Cox proportional hazard model with covariates using the first PLS scores as covariates.
SurvPlsClass(Survival, Mdata, Censor, Reduce = TRUE, Select = 150, Prognostic = NULL, Plots = FALSE, Quantile = 0.5)SurvPlsClass(Survival, Mdata, Censor, Reduce = TRUE, Select = 150, Prognostic = NULL, Plots = FALSE, Quantile = 0.5)
Survival |
A vector of survival time with length equals to number of subjects |
Mdata |
A large or small metabolic profile matrix. A matrix with metabolic profiles where the number of rows should be equal to the number of metabolites and number of columns should be equal to number of patients. |
Censor |
A vector of censoring indicator |
Reduce |
A boolean parameter indicating if the metabolic profile matrix should be reduced, default is TRUE and larger metabolic profile matrix is reduced by supervised pca approach and first pca is extracted from the reduced matrix to be used in the classifier. |
Select |
Number of metabolites (default is 15) to be selected from supervised PCA. This is valid only if th argument Reduce=TRUE |
Prognostic |
A dataframe containing possible prognostic(s) factor and/or treatment effect to be used in the model. |
Plots |
A boolean parameter indicating if the plots should be shown. Default is FALSE |
Quantile |
The cut off value for the classifier, default is the median cutoff |
This function reduces larger metabolomics matrix to smaller version using supervised pca approach. The function performs the PLS on the reduced metabolomics matrix and fit Cox proportional hazard model with first PLS scores as a covariate afterwards. And classifier is then built based on the first PLS scores multiplied by its estimated regression coefficient. Patients are classified using median of the risk scores. The function can also perform grid analysis where the grid will be several quantiles but the default is median. This function can handle single and multiple metabolites. Prognostic factors can also be included to enhance classification.
A object is returned with the following values
Survfit |
The cox proportional regression result using the first PCA |
Riskscores |
A vector of risk scores which is equal to the number of patents. |
Riskgroup |
The classification of the subjects based on the PCA into low or high risk group |
pc1 |
The First PCA scores based on either the reduced Metabolite matrix or the full matrix |
KMplot |
The Kaplan-Meier survival plot of the riskgroup |
SurvBPlot |
The distribution of the survival in the riskgroup |
Riskpca |
The plot of Risk scores vs first PCA |
Olajumoke Evangelina Owokotomo, [email protected]
Ziv Shkedy
Bair E, Hastie T, Debashis P, Tibshirani R (2006). “Prediction by supervised principal components.” American Statistics Association,, 101(473), 119-137.
coxph,
EstimateHR, plsr,
SurvPcaClass
## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Result = SurvPlsClass(Survival=Data$Survival, Mdata=t(Data$Mdata), Censor=Data$Censor, Reduce = FALSE, Select = 150, Prognostic = Data$Prognostic, Plots = FALSE, Quantile = 0.5) ## GETTING THE SURVIVAL REGRESSION OUTPUT Result$SurvFit ## GETTING THE RISKSCORES Result$Riskscores ### GETTING THE RISKGROUP Result$Riskgroup ### OBTAINING THE FIRST PRINCIPAL COMPONENT SCORES Result$pc1## FIRSTLY SIMULATING A METABOLIC SURVIVAL DATA Data = MSData(nPatients = 100, nMet = 150, Prop = 0.5) ## USING THE FUNCTION Result = SurvPlsClass(Survival=Data$Survival, Mdata=t(Data$Mdata), Censor=Data$Censor, Reduce = FALSE, Select = 150, Prognostic = Data$Prognostic, Plots = FALSE, Quantile = 0.5) ## GETTING THE SURVIVAL REGRESSION OUTPUT Result$SurvFit ## GETTING THE RISKSCORES Result$Riskscores ### GETTING THE RISKGROUP Result$Riskgroup ### OBTAINING THE FIRST PRINCIPAL COMPONENT SCORES Result$pc1