braidrm (version 1.0.3)

evalFlippedBraidModel: Evaluate Flipped BRAID Surfaces

Description

Evaluate Flipped BRAID Surfaces

Usage

evalFlippedBraidModel(DA, DB, bpar, flip)

Value

A numeric vector the same length as DA and/or DB with the predicted flipped BRAID response surface values.

Arguments

DA

A vector of concentrations of drug A in a combination (values 0 and Inf are permitted). Must be length 1 or the same length as DB.

DB

A vector of concentrations of drug B in a combination (values 0 and Inf are permitted). Must be length 1 or the same length as DA.

bpar

Flipped-BRAID parameter of the flipped response surface. See details for more information on specifying atypical surfaces

flip

String specifying the direction or directions of the surface's flip. Must be one of "A", "B", or "both".

Details

While the BRAID model generates a fairly versatile range of combined behaviors, the traditional model is still strictly constrained in certain respects. Response surfaces must exhibit a change in response to both drugs and this change must be in the same direction. Furthermore, the model is only suited to surfaces in which the change resulting from single drugs is larger than the additional effect of the combination. However, modifying the model equation by inverting one or both of the slope parameters can produce a set of qualitatively distinct "flipped" surfaces, allowing the model to produce a much wider range of behaviors. For example, flipping the model along the axis representing drug A can produce a surface in which drug B has no effect in isolation, but attenuates or eliminates the effect of drug A, a pattern we call a "protective" surface. See fitBraidFlipped() for the possible range of surfaces.

An important note: a flipped BRAID surface, like a traditional BRAID surface, is represented by a parameter vector of up to 9 values. While these functions will attempt to fill a 7- or 8- value parameter factor to a full 9-element vector, it is strongly recommended that you specify the response surface with the full 9-element vector, as the precise ordering with which implicit values are arranged can be extremely confusing. Note also that flipped parameter vectors, regardless of the underlying mathematical representation, should always be specified in the same order as in a traditional vector. So in a full 9-element vector:

• Parameter 6 (E0) should always specify the expected effect when both drugs are absent

• Parameter 7 (EfA) should always specify the expected effect at high concentrations of drug A when drug B is absent

• Parameter 8 (EfB) should always specify the expected effect at high concentrations of drug B when drug A is absent

• Parameter 9 (Ef) should always specify the expected effect when both drugs are present at high concentrations

Examples

concentrations <- c(0, 2^(-3:3))
surface <- data.frame(
    concA = rep(concentrations,each=length(concentrations)),
    concB = rep(concentrations,times=length(concentrations))
)

surface$protective <- evalFlippedBraidModel(
    surface$concA,
    surface$concB,
    c(1, 1, 3, 3, 0, 0, 100, 0, 10),
    flip="A"
)
surface$coactive <- evalFlippedBraidModel(
    surface$concA,
    surface$concB,
    c(1, 1, 3, 3, 0, 0, 0, 0, 100),
    flip="both"
)

head(surface)