Two-way fitting for projected dose-response curve.

Fit projected dose-response curves with E_min as the viability of the treatment being added to the other treament at a fixed dose.

Usage

fitTwowayZIP(
  combo_profiles,
  residual = "logcosh",
  show_Rsqr = TRUE,
  nthread = 1L,
  optim_only = TRUE,
  loss_args = list()
)

Arguments

combo_profiles

data.table contains three parameters of dose-response curves for each single agent in a drug comnbination, and the observed viability of two treatments combined.

residual

character Method used to minimise residual in fitting curves. 3 methods available: c("logcosh", "normal", "Cauchy"). The default method is logcosh. It minimises the logarithmic hyperbolic cosine loss of the residuals and provides the fastest estimation among the three methods, with fitting quality in between normal and Cauchy; recommanded when fitting large-scale datasets. The other two methods minimise residuals by considering the truncated probability distribution (as in their names) for the residual. Cauchy provides the best fitting quality but also takes the longest to run.

show_Rsqr

logical whether to show goodness-of-fit value in the result.

nthread

integer Number of cores used to perform computation. Default 1.

optim_only

logical(1) Should the fall back methods when optim fails

loss_args

list Additional argument to the loss function. These get passed to losss via do.call analagously to using ....

Value

data.table contains parameters of projected dose-response curves for adding one treatment to the other.

References

Yadav, B., Wennerberg, K., Aittokallio, T., & Tang, J. (2015). Searching for Drug Synergy in Complex Dose–Response Landscapes Using an Interaction Potency Model. Computational and Structural Biotechnology Journal, 13, 504–513. https://doi.org/10.1016/j.csbj.2015.09.001

Examples

if (FALSE) {
combo_profiles <- CoreGx::buildComboProfiles(tre, c("HS", "EC50", "E_inf", "viability"))
combo_twowayFit <- fitTwowayZIP(combo_profiles)
}