R/ENSEMBLE.R
SCDC_ENSEMBLE_subcl.RdENSEMBLE function for manually input deconvolution results
SCDC_ENSEMBLE_subcl(bulk.eset, prop.list, truep = NULL, ct.sub, search.length = 0.05, search.method = c("Grid search", "LAD", "NNLS"), iter.max = 2000, nu = 1e-04, epsilon = 0.001, ...)
| bulk.eset | ExpressionSet object for bulk samples |
|---|---|
| prop.list | list of deconvolution results. should include: yhat (genes by samples), prop.est(samples by cell types). It's recommended to use as many genes as possible. |
| truep | true cell-type proportions for bulk samples if known |
| ct.sub | a subset of cell types that are selected to construct basis matrix |
| search.length | if using "Grid search", the step length used. |
| search.method | method to derive the ENSEMBLE weights, including "Grid search", "LAD", "NNLS". |
| iter.max | the maximum number of iteration in WNNLS |
| nu | a small constant to facilitate the calculation of variance |
| epsilon | a small constant number used for convergence criteria |
| ct.varname | variable name for 'cell types' |
| sample | variable name for subject/samples |