I'm using the pRoloc package to train an TAGM-MCMC model on a proteome data set. For this a Markov Chain Monte Carlo simulation is used via the
tagmMcmcTrain() function. However, if I try to run the simulation, an error message occurs:
> tagm_params <- tagmMcmcTrain(data, fcol = "localization", numIter = 15000, burnin = 4000, thin = 10, numChains = 4) error: chol(): decomposition failed Fehler: BiocParallel errors element index: 1, 2, 3, 4 first error: chol(): decomposition failed
I understand, that the problem roots in the covariance matrix not being positive definite. I confirmed this by calculating the eigenvalues of the covariance Matrix. I can correct this by calculating the nearest positive definite covariance matrix by using the
nearPD() function. But how can I pass this corrected covariance matrix to the
Thank you very much in advance!