run_scINSIGHT {scINSIGHT}R Documentation

Perform scINSIGHT on normalized datasets

Description

Perform INterpreting single cell gene expresSIon bioloGically Heterogeneous daTa (scINSIGHT) to return factorized W1W_{\ell1}, W2W_{\ell2}, HH and VV matrices.

This factorization produces a W1W_{\ell1} matrix (cells by KjK_j), a W2W_{\ell2} matrix (cells by KK), a shared VV matrix (KK by genes) for each sample, and a HH (KjK_j by genes) matrix for each condition. W2W_{\ell2} are the expression matrices of KK common gene modules for all samples, VV is the membership matrix of KK common gene modules, and it's shared by all samples. W1W_{\ell1} are the expression matrices of KjK_j condition-specific gene modules for all samples, and HH are the membership matrices of KjK_j condition-specific gene modules for all conditions.

Usage

run_scINSIGHT(
  object,
  K = seq(5, 15, 2),
  K_j = 2,
  LDA = c(0.001, 0.01, 0.1, 1, 10),
  thre.niter = 500,
  thre.delta = 0.01,
  num.cores = 1,
  B = 5,
  out.dir = NULL,
  method = "increase"
)

Arguments

object

scINSIGHT object.

K

Number of common gene modules. (default c(5, 7, 9, 11, 13, 15))

K_j

Number of dataset-specific gene modules. (default 2)

LDA

Regularization parameters. (default c(0.001, 0.01, 0.1, 1, 10))

thre.niter

Maximum number of block coordinate descent iterations to perform. (default 500)

thre.delta

Stop iteration when the reduction of objective function is less than the threshold. (default 0.01)

num.cores

Number of cores used for optimizing factorizations in parallel (default 1).

B

Number of repeats with random seed from 1 to B. (default 5)

out.dir

Output directory of scINSIGHT results. (default NULL)

method

Method of updating the factorization (default "increase"). If provide multiple KK, user can choose method between "increase" and "decrease".

For "increase", the algorithm will first perform factorization with the least K=K1K=K_1. Then initialize K2K1K_2-K_1 facotrs, where K2K_2 is the KK sightly larger than K1K_1, and perform facotrization with these new facotrs. Continue this process until the largest KK.

For "increase", the algorithm will first perform factorization with the largest K=K1K=K_1. Then choose K2K_2 facotrs, where K2K_2 is the KK sightly less than K1K_1, and perform facotrization with these new facotrs. Continue this process until the least KK.

Value

scINSIGHT object with W1W_1, W2W_2, HH, VV and parameters slots set.


[Package scINSIGHT version 0.1.4 Index]