search

Wedge-lab / dpclust

Dirichlet Process based methods for subclonal reconstruction of tumours
GNU Affero General Public License v3.0
28 stars 16 forks source link

memory usage issue #7

Open bioinfograd opened 4 years ago

bioinfograd commented 4 years ago

Hello, I have 11 samples from 1 patient so I am trying to run an nd analysis. I ran this successfully on 3 samples but when I increase, when "Estimating density for all MCMC iterations..." the vector being made is gigantic. This is being run on 300 mutations. Is there any way to reduce this or bypass this? I would love to get the the mutation clusters and CCF of each for each sample for the 1 patient.

Running all 11 samples error message: Error in array(NA, c(gridsize, length(sampledIters))) : vector is too large Calls: RunDP ... multiDimensionalClustering -> Gibbs.subclone.density.est.nd -> array

Running all 10 samples error message: Error: cannot allocate vector of size 13301026.9 Gb

Running all 9 samples error message: Error: cannot allocate vector of size 738945.9 Gb

Any advise you can give would be greatly appreciated.

bioinfograd commented 4 years ago

After going through it more, it would be helpful to add args for resolution and maxburden at the start when running dpclust_pipeline.R.