Deconvolution of Tumour Profiles
Perform calculation for Tumour Adjacent Cell (TAC) profiles
Minimize a differentiable multivariate function
Compute the derivative of the loglikelihood relevant to vv for step 1
Compute the loglikelihood relevant to vv for step 1
Perform first step of ISOpure purification algorithm
Perform second step of ISOpure purification algorithm
Log-sum-exp
Greater than operator
Less than operator
Modified logarithm function
Tiles matrix horizontally or vertically
Compute loglikelihood given all model parameters for step 1
Initialize a model list to hold all the parameters
Optimizes the ISOpure parameters for step 1
Compute loglikelihood relevant to kappa for step 1
Compute derivative of loglikelihood with respect to kappa for step 1
Compute loglikelihood relevant to kappa for step 1
Compute the derivative of the loglikelihood relevant to mm for step 1
Compute the loglikelihood relevant to mm for step 1
Compute loglikelihood relevant to omega for step 1
Compute the derivative of loglikelihood relevant to omega for step 1
Compute the loglikelihood relevant to omega for step 1
Optimize kappa in step 1
Optimize the reference cancer profile, m, in step 1
Optimize omega in step 1
Optimize theta in step 1
Optimize vv in step 1
Compute the derivative of loglikelihood relevant to theta for step 1
Compute the loglikelihood relevant to theta for step 1
Compute loglikelihood relevant to vv for step 1
Compute loglikelihood given all model parameters for step 2
Compute loglikelihood given all model parameters for step 2
Optimizes the ISOpure parameters for step 2
Compute the derivative of loglikelihood relevant to the patient cancer...
Compute the loglikelihood relevant to the patient cancer profiles, cc,...
Compute loglikelihood relevant to kappa for step 2
Compute derivative of loglikelihood with respect to kappa for step 2
Compute loglikelihood relevant to kappa for step 2
Optimize the tumor-specific cancer profiles in step 2
Optimize kappa in step 2
Optimize theta in step 2
Optimize vv in step 2
Compute the derivative of loglikelihood relevant to theta for step 2
Compute the loglikelihood relevant to theta for step 2
Compute loglikelihood relevant to vv for step 2
Deconvolution of mixed tumour profiles into normal and cancer for each patient, using the ISOpure algorithm in Quon et al. Genome Medicine, 2013 5:29. Deconvolution requires mixed tumour profiles and a set of unmatched "basis" normal profiles.