createParameterMatrix: Create or Modify the SpeCond argument parameters

Description

createParameterMatrix creates and/or modifies the param.detection matrix used as argument in the SpeCond function. If parm.detection is NULL the param.detection matrix used is the one containing the default parameter values, as obtained by getDefaultParameter. The remaining arguments enable to change the values of the param.detection matrix.

Usage

createParameterMatrix(param.detection = NULL, beta.1 = NULL, beta.2 = NULL,
lambda.1 = NULL, lambda.2 = NULL, per.1 = NULL, per.2 = NULL, md.1 = NULL, 
md.2 = NULL, mlk.1 = NULL, mlk.2 = NULL, rsd.1 = NULL, rsd.2 = NULL, 
pv.1 = NULL, pv.2 = NULL)

Arguments

param.detection

a matrix of 2 rows and 7 columns as the result of getDefaultParameter

beta.1

Influences the prior applied during the determination of the variance of the normal distributions. It is necessary in the first fitting step to allow the model to capture isolated outliers.

beta.2

The normal use of SpeCond does not prior on Step2: must be set to 0

lambda.1

Influences the choice of models by affecting the selection of one, two or three normal distributions, thus introducing some weight on the effect of number of parameters to be defined. The default is 1, the model uses the BIC value taking into account the log-likelihood value.

lambda.2

Same as lambda.1 for the second step of the SpeCond function

per.1

percentage threshold: this is the percentage of conditions that can be detected as specific. As per increases a larger number of expression values per genes can be identified as specific. The default is 0.1

per.2

percentage threshold: This is the final percentage of condition that can be detected as specific. As per increases a larger number of expression values per genes can be identified as specific. The default is 0.3

md.1

median difference: this is the minimum value between the median values of two mixture components that is allowed to identify one of them as representing outliers, i.e. possibly not part of the null distribution. This corresponds to a biological fact; specific expression that corresponds to noise should not be detected as specific

md.2

Same as md.1 for the second step of the SpeCond function. For consistency should be equal to md.1

mlk.1

minimum log-likelihood: enables the identification of clusters of conditions that are well separated from the others in the model. If the gene mlk value>mlk, the mixture component can be detected as outlier (i.e. not part of the null distribution)

mlk.2

same as mlk.2 for the second step of the SpeCond function

rsd.1

minimum of standard deviation ratio: enables the identification of clusters of conditions that are extremely spread out compared to the distribution clustering of most expression values. If the gene rsd values< rsd the mixture component can be detected as outlier (i.e. not part of the null distribution)

rsd.2

same as rsd.1 for the second step of the SpeCond function

pv.1

p-value threshold to detect a condition as specific for a given gene

pv.2

same as pv.1 for the second step of the SpeCond function, for consistency should be equal to pv.1

Value

param.detection: a matrix of 2 row and 7 columns. The rows "Step1 "and "Step2" correspond respectively to the first and second set of parameters for the SpeCond function. The parameters (columns) are: lambda, beta, per, md, mlk, rsd. See the createParameterMatrix documentation for more details about the parameters.

Warning

The SpeCond code is based on: beta.2=0 md.1=md.2 per.1

Examples

Run this code

##Get the default parameters and changing the mlk.1 value to 10:
param.detection2=createParameterMatrix(mlk.1=10)
param.detection2
## Modify param.detection2 with mlk.1 value to 15 and rsd.2 value to 0.2
param.detection2B=createParameterMatrix(param.detection=param.detection2,
mlk.1=10, rsd.2=0.2) 
param.detection2B