calcVecLMs {CNVScope}

Create a linear regression matrix.

Description

Creates a matrix of linear regression p-values, log transformed from every combination of columns in the parent matrix.

Usage

calcVecLMs(
  bin_data,
  use_slurm = F,
  job_finished = F,
  slurmjob = NULL,
  n_nodes = NULL,
  cpus_on_each_node = 2,
  memory_per_node = "2g",
  walltime = "4:00:00",
  partitions = "ccr,quick"
)

Arguments

bin_data

The parent matrix, with columns to have linear regression performed on them.

use_slurm

Paralleize over a number of slurm HPC jobs? If false, the program will simply run locally.

job_finished

Are all the slurm jobs finished and the results need retrieving?

slurmjob

the slurm job object produced by rslurm::slurm_apply(), after running the function initially.

n_nodes

the number of nodes used in your slurm job.

cpus_on_each_node

The number of cpus used on each node

memory_per_node

the amount of ram per node (e.g. "32g" or "2g")

walltime

Time for job to be completed for SLURM scheduler in hh:mm:ss format. Defaults to 4h.

partitions

the partitions to which the jobs are to be scheduled, in order of priority.

Value

The output matrix, or if using slurm, the slurm job object (which should be saved as an rds file and reloaded when creating the output matrix).

Examples

#small example
#bin_data<-matrix(runif(5*5),ncol=5)
foreach::registerDoSEQ()
#full_matrix<-suppressWarnings(calcVecLMs(bin_data))
#Please note that lm() will make a warning when there are two vectors that are too close 
#numerically (this will always happen along the diagonal).
#This is normal behavior and is controlled & accounted for using this function as well as
#the postProcessLinRegMatrix function (which converts the infinite values to a maximum).