designScore: Calculate the A- or D- optimality score based on current...

Description Usage Arguments Details Value Author(s) References See Also


According to the current experimental design, the Fisher information matrix is obtained and then either the A- or D- optimality score is computed.


  designScore( genotype, array.allocation, condition.allocation, 
               nEnvFactors, nLevels, Level, nConditions, weight=1, 
               optimality="A", bTwoColorArray, envFactorNames)



genotype data: a nMarker-by-nRILs matrix with two allels being 0 and 1 (or A and B) or three allels being 0, 0.5 and 1 (or, A, H, and B), where 0.5 (or H) represents heterozygous allele.


matrix with nArray rows and nRIL columns. Elements of 1/0 indicate this RIL (or strains) is/not selected for this array.


matrix with nCondition rows and nRIL columns. Elements of 1/0 indicate this RIL (or strains) is/not selected for this condition.


number of environmental factors, an integer bewteen 1 and 3. When nEnvFactors is 1 and the number of levels for the enviromental factor (nLevels)is 1, there is one condition in the experiment (i.e. no enviromental perturbation) and thus only genetic factor will be considered in the algorithm. When nEnvFactors is 1 and nLevels is larger than 1 or nEnvFactors is larger than 1, all main factor(s) and interacting facotr(s) will be included. Examples: If there is a temperature perturbation, then nEnvFactors is 1; If there is both temperature and drug treatment perturbation, then nEnvFactors is 2.


number of levels for each factor, a vector with each component being an integer. The length of it should equal nEnvFactors.


a list which specifies the levels for each factor in the experiment. There are in total nEnvFactors elements in the list and each element correpsond to certain envrironmental factor. The emlemet is a vector describing all levels of the environmental factor. default setting for the level of each factor is 1, 2, ... nLevels[i]. (Here nLevels[i] is the ith element of nLevels, which gives the total number of levels for i environmental factor).


number of all possible combination of all environmental factors.


a vector with length of variableNumber which is calculated from function variableNumber. Default = 1 (which means the parameters to be estimated are equally important during optimization.)


type of optimality, i.e. "A" (A-optimality) or "D" (D-optimality). A-optimality minimizes $Trace((X'X)^-1)$, which corresponds to minimum average variance of the parameter estimates. D-optimality minimizes $det(X'X)^-1$, which corresponds to minimum generalized variance of the parameter estimates.


binary variable indicating experiment type:
bTwoColorArray <- TRUE \#for dual channel experiment
bTwoColorArray <- FALSE \#for single channel experiment


a vector with names for all environmental factor(s). For example, for the experiment with two environmental factors of temperature and drug treatment: envFactorNames <- c( "Temperature", "Dosage" )
Default = NULL, then the output will use "F1" and "F2" to indicate the environmental factors.


Example parameter settings:
Suppose to design an experiment with two environmental factors (F1, F2) and there are two diffferent levels for each environment. The levels are 16 and 24 for F1, and 5 anf 10 for F2. Thus the following command can be used:
nEnvFactors <- 2
nLevels <- c ( 2, 2 )
levels <- list ( c(16, 24), c(5, 10) )
The length of parameter weight is dependent on the number of environmental factors:
When nEnvFactor = 0,
weight is 1 as there is only one parameter of interest (genotype).
When nEnvFactor = 1,
weight = c( $w_Q$, $w_F1$, $w_QF1$ )
When nEnvFactor = 2,
weight = c( $w_Q$, $w_F1$, $w_F2$, $w_QF1$, $w_QF2$, $w_F1F2$, $w_QF1F2$)
When nEnvFactor = 3,
weight = c( $w_Q$, $w_F1$, $w_F2$, $w_F2$, $w_QF1$, $w_QF2$, $w_QF3$, $w_F1F2$, $w_F1F3$, $w_F2F3$, $w_QF1F2$, $w_QF1F3$, $w_QF2F3$, $w_QF1F2F3$ )
Here $w_Q$ represents the weight for genotype effect, $w_F1$ represent the weight for F1 effect and $w_QF1$ represent the weight for interaction between genotype and F1 effect, etc.


The score is defined as the "double" sum of the variances, summed over all parameters and over all markers.


Yang Li <>, Gonzalo Vera <>
Rainer Breitling <>, Ritsert Jansen <>


Y. Li, M. Swertz, G. Vera, J. Fu, R. Breitling, and R.C. Jansen. designGG: An R-package and Web tool for the optimal design of genetical genomics experiments. BMC Bioinformatics 10:188(2009)
Y. Li, R. Breitling and R.C. Jansen. Generalizing genetical genomics: the added value from environmental perturbation, Trends Genet (2008) 24:518-524.
E. Wit and J. McClure. Statistics for Microarrays: Design, Analysis and Inference. (2004) Chichester: Wiley.

See Also


Search within the designGG package
Search all R packages, documentation and source code

Questions? Problems? Suggestions? or email at

Please suggest features or report bugs with the GitHub issue tracker.

All documentation is copyright its authors; we didn't write any of that.