Parameter estimation and sensitivity analysis framework for colon cancer (An NCI-NIH funded project)
Codes for parameter estimation and sensitivity analysis of QSP models for colon cancer. Though the current model is a modification of the one given in the paper by dePillis et. al. (2014), "Mathematical Model of Colorectal Cancer with Monoclonal Antibody Treatments", doi:10.9734/BJMMR/2014/8393, the codes are generic as they can be appropriately modified for any QSP model.
This is a part of the National Cancer Institute-NIH funded project titled "Data-driven QSP software for personalized colon cancer treatment" Grant number: R21 CA242933-01
Developers: Achyuth Manoj, Susanth Kakarla, Juan Villegas, Suvra Pal and Souvik Roy.
PARAMETER ESTIMATION-
DRIVER.m: The main file to run a optimization algorithm for estimating parameters.
forward.m: Solves the forward ODE QSP model for colon cancer. Should modify if the QSP model changes.
adjoint.m: solves the adjoint ODE. Should modify if the QSP model changes.
data.m: Generates the patient data to be fed in as input for the optimization setup. Should modify if the QSP model changes.
J.m: Functional to be minimized that includes the data-fitting least squares terms and regularization terms. Should modify if the QSP model changes.
gradient.m: Computes the gradient of the reduced functional. Should modify if the QSP model changes.
optim.m: Optimization solver.
lin_search.m: Line search algorithm required for the gradient step of the optimization solver.
inner_g.m: Computes the inner product of 2 vectors.
parameters.m: Contains the list of all the known and user defined parameter values for the colon cancer model. Should modify if the QSP model changes.
SENSITIVTY ANALYSIS-
LHS.m: Computes the Latin hypercube samples (LHS) of the uncertain parameters obtained from the optimization step.
Weibull_par.m: Computes the LHS for a given uncertain parameter using a Weibull distribution.
Newton.m: Computes the scale and the shape parameter of a Weibull distribution in a given interval.
par_corr.m: Computes the PRCC of the uncertain parameters with respect to the tumor cell count.
The pdf file is the summary of the actual paper under preparation based on which the codes have been developed.