DeCompress pipeline functions

Functions for the main pipeline

findInformSet()

Select the compartment specific genes

trainCS()

Wrapper to train compression matrix

bestDeconvolution()

Find cell-type gene signatures and do deconvolution

Estimate number of compartments

Functions for data-driven methods of estimating number of compartments

kaiser()

Kaiser method to determine number of important SVs

findNumberCells()

Select the number of cell types using SVD methods

Feature selection

Functions for Step 1 feature selection

vardecomp()

Select CTS genes with variance proporties

Train compressed sensing

Functions for predictive modeling

TVMagic()

Elastic-net for compressed sensing

bigstatsenet()

Elastic-net for compressed sensing using bigstatsr

l1Magic()

Non-linear for compressed sensing

l2Magic()

Elastic-net for compressed sensing

lar()

Least angle regression for compressed sensing

trainCS_gene()

Wrapper to train compression matrix

Deconvolution

Functions for deconvolution

csDeCompress()

TOAST-NMF deconvolution

estimateUnmix()

Run unmix from DESeq2

linCor()

Use mutual linearity to find cell-type gene signatures and do deconvolution

nmfOut()

NMF function for TOAST

iterateSigs()

Use NMF repeatedly with TOAST framework to find cell-type gene signatures and do deconvolution

Miscellaneous

Functions for calculating error or expanding target

calculateError()

Find errors in estimated proportion matrix and signature

mseError()

Select the number of cell types using SVD methods

expandTarget()

Expand targetted panel to larger feature space