runBRAIDanalysis {braidrm}

Run Full BRAID Analysis

Description

Performs a complete analysis of combined action data using the BRAID model. The behaviors of both drugs alone (if such doses are included) are fit using a Hill equation; if desired, the resulting Hill equation fits can be used to correct uncertain or noisy input concentrations for use in combination analysis. All measurements of the two drugs (in isolation or in combination) are then fit using the BRAID equation.

Usage

runBRAIDanalysis(data, defaults, llims = NULL, ulims = NULL, itype = 1,
					compounds = NULL, corrconc = FALSE, corrsigr = 1)

Arguments

data

a data frame containing all measurements to be fit. Must conatin columns named "conc1", "conc2", and "act"; other columns may be required (see Details below).

defaults

a two-element vector containing default values for the initial and maximal effects of the combination; used by findBestBRAID to select best BRAID model

llims

a ten-element vector of lower limits on parameters being fit. Any parameters that do not require a limit can have a value of NA.

ulims

a vector of upper limits on parameters being fit. Follows same behavior as llims.

itype

an integer that specifies the type of interaction(s) that is assumed in the models. Used by findBestBRAID

compounds

an optional parameter specifying which compounds are to be fit; this requires that compounds be identified in the input data frame data.

corrconc

boolean specifying whether input concentrations will be corrected according to individual dose response using hillConcCorrect

corrsigr

the ratio of noise in measurement to noise the base-10 logarithm of concentration; used by hillConcCorrect if corrconc is TRUE

Details

This convenience function is intended as a blueprint for a complete BRAID model analysis. Though users of this package can develop their own approaches to applying the BRAID model to combined action data (indeed we encourage it), this function encapsulates our overall strategy for fitting the BRAID equation. The input parameter data must contain all measurements to be fit, with the columns "conc1", "conc2", and "act" containing the corresponding inputs and outputs. If no other columns required (based on the values of other parameters), "conc1" is assumed to describe the concentration of drug A, and "conc2" the concentration of drug B, with values of 0 for "conc1" or "conc2" corresponding to conditions with Drug B or Drug A alone, respectively.

If data contains measurements from multiple combinations, the identity of the compounds to be analyzed must be specified in the parameter compounds. Further, the input data frame data must contain columns named "compound1" and "compound2" that specify which two compounds were used in each measurement. Measurements will be used in which "compound1" is equal to the first element of compounds or "conc1" is equal to 0, and in which "compound2" is equal to the second element of compounds or "conc2 is equal to 0. Note that this means measurements in which "conc1" and "conc2" are both equal to 0 will be used in any analysis.

For concentrations to be corrected in the combination analysis, there must be a correspondence between concentrations of each drug alone and concentrations used in combination; in our implementation, this is assumed to be the result of a shared drugging plate, such that measurements corresponding to the same well are given the same actual concentration of drug. As a result, if corrconc is set to TRUE, each measurement must be associated with a well specified in a column named "well".

Value

A list with the following componenents:

concs

A two-column array containing the concentrations of the first and second drugs used in the analysis. This will only inlcude measurements from the input frame data that contained drug A alone, drug B alone, or drug A and B in combination. Even if corrconc is TRUE, this array will contain the original, uncorrected concentrations. To access the corrected concentrations, use the conc1 and conc2 elements of the component braidFit (described below).

act

The measured response values corresponding to the concentration pairs in concs.

corrconc

The value of the input parameter corrconc.

corrsigr

The value of the input parameter corrsigr.

braidFit

The best BRAID model fit as determined by findBestBRAID.

hfit1

A set of Hill model fits to the behavior of drug A in isolation, as output by findBestHill. If no data points for drug A in isolation are included, this component is NULL.

hfit2

A set of Hill model fits to the behavior of drug B in isolation, as output by findBestHill. If no data points for drug B in isolation are included, this component is NULL.

Author(s)

Nathaniel R. Twarog

See Also

braidrm, findBestBRAID, findBestHill, hillConcCorrect

Examples

data(es8olatmz)
# Note that 'es8olatmz' contains all necessary column names, including
# 'conc1','conc2','act','compound1','compound2', and, for concentration
# correction, 'well'
## Not run: brdAnalysis <- runBRAIDanalysis(es8olatmz,defaults=c(0,-2.7),corrconc=TRUE)
brdAnalysis <- runBRAIDanalysis(es8olatmz,defaults=c(0,-4))
summary(brdAnalysis$braidFit)