NMF_RNA_Tang_ind: NMF core function

In tdw1221/NITUMID: Non-negative Matrix Factorization-based Immune-TUmor MIcroenvironment Deconvolution

Description

'NMF_RNA_Tang_ind' is a upper level wrapper function for 'NMF_RNA_Tang'. On top of core NMF, it also initialtes $W$ and $H$, and scales/normalizes according to the input arguments: scale_Data and reg_para. Besides, it also does random permutations to check results stability and helps choose the best parameters. Although it has full sets of NITUMID's functions, the 'NMF_RNA_Tang_ind' is still not supposed to be run independently, has hyper-function will call it with different parameters choices.

Usage

NMF_RNA_Tang_ind(Y, lam_W, beta_W, A, eta, tol = 10^-5, prterr = F,
  max.itr = 2000, always = F, num_cell = 11, scale_Data,
  perm_type = T)

Arguments

Y	Y is the gene expression matrix that has #signature genes rows and #samples columns
lam_W	$\lambda$ is the parameter imposing A's structure into resulting W, see our paper for more info. In real calculation, upper level functions will automatically choose suitable $\lambda$ no need to set by yourself
beta_W	$\beta$ is the parameter controlling sparsity ridge penalty, see our paper for more information. In real calculation, upper level functions will automatically choose suitable $\beta$ no need to set by yourself
A	The trichotomous guide matrix A
eta	$\eta$ is the parameter for ensuring column sums of H are 1s, upper level function will choose suitable $\eta$ automatically, no need to spcify yourself
tol	numeric scalar, the cutoff value for stop iterating, default $10^-5$
prterr	logical value indicating if it should print each step's error, default FALSE
max.itr	numerical scalar, maximum iteration number, default 2000
num_cell	numeric scalar, number of component cell types, dfault 11. Noted that if you were to change component cell types number, you should change your $A$ matrix accordingly.
scale_Data	the parameter controlling which scaling method to use for $Y$: 0 indicates no scaling, 1 indicates scaling $||Y||_F=\sqrtnm$, where $n,m$ are dimensions of Y; 2 indicates scaling such that $||Y^scaled||_F=||log(Y^raw+1)||_F$
perm_type	logical variable, default TRUE. If TRUE, the function will do a permutation and measure consistency between results.
ini.W	Initial W matrix
ini.H	Initial H matrx
alaways	logical, if it should output results whatsover when max.itr is reached, default TRUE

Value

'NMF_RNA_Tang' returns the resulted W and H, as well as track of errors along iteration

Examples

This function is not supposed to be run independently, it is supposed to be called by top level function, which will feed it with different pre-determined
parameters and choose thr best one

tdw1221/NITUMID documentation built on May 31, 2019, 1:54 a.m.