Computes the derivative of the part of the loglikelihood function relevant to optimizing kappa for step 1. Instead of performing constrained optimization on kappa directly, we optimize the log of kappa in an unconstrained fashion. Thus, if y=log(kappa) and L is the loglikelihood function w.r.t. y, to optimize L w.r.t. y, dL/dy = dL/dkappa * dkappa/dy, where dkappa/dy = exp(y)= exp( log(kappa)). The input into the derivative function is log(kappa - model\$MIN_KAPPA).
ISOpureS1.model_optimize.kappa.kappa_deriv_loglikelihood(log_kappa, tumordata, model)the scalar log(kappa - model\$MIN_KAPPA)
a GxD matrix representing gene expression profiles of tumour samples
list containing all the parameters to be optimized
The negative derivative of the part of the loglikelihood function relevant to kappa with respect to log kappa (a scalar given that for step 1 of ISOpure kappa is a scalar)