gsdecon
C++ port of the GSDecon algorithm
|
Options for compute()
and compute_blocked()
.
More...
#include <Options.hpp>
Public Attributes | |
int | rank = 1 |
bool | scale = false |
scran_blocks::WeightPolicy | block_weight_policy = scran_blocks::WeightPolicy::VARIABLE |
scran_blocks::VariableWeightParameters | variable_block_weight_parameters |
int | num_threads = 1 |
bool | realize_matrix = true |
irlba::Options | irlba_options |
Options for compute()
and compute_blocked()
.
scran_blocks::WeightPolicy gsdecon::Options::block_weight_policy = scran_blocks::WeightPolicy::VARIABLE |
Policy to use for weighting batches of different size, for compute_blocked()
.
irlba::Options gsdecon::Options::irlba_options |
Further options to pass to irlba::compute()
.
int gsdecon::Options::num_threads = 1 |
Number of threads to use. The parallelization scheme is determined by tatami::parallelize()
and irlba::parallelize()
.
int gsdecon::Options::rank = 1 |
Rank of the low-rank approximation. Higher values can capture more biological signal at the risk of including more noise. The default value of 1 assumes that each gene set only describes a single coordinated biological function.
bool gsdecon::Options::realize_matrix = true |
Whether to realize tatami::Matrix
objects into an appropriate in-memory format before PCA. This is typically faster but increases memory usage.
bool gsdecon::Options::scale = false |
Should genes be scaled to unit variance? Genes with zero variance are ignored. This ensures that each gene contributes equally to the PCA, favoring consistent variation across many genes rather than large variation in a few genes.
scran_blocks::VariableWeightParameters gsdecon::Options::variable_block_weight_parameters |
Parameters for the variable block weights for compute_blocked()
. Only used when Options::block_weight_policy = scran_blocks::WeightPolicy::VARIABLE
.