EM: EM

In lch14forever/BEEM: BEEM: Estimating Lotka-Volterra models from time-course microbiome sequencing data

EM

Description

Function to estimate biomass and parameters simultaneously

Usage

EM(
  dat,
  meta,
  forceBreak = NULL,
  useSpline = TRUE,
  dev = 5,
  verbose = TRUE,
  refSp = NULL,
  warmup_iter = min_iter/2,
  min_iter = 30,
  max_iter = 100,
  converge_thres = 0.001,
  ncpu = 2,
  seed = NULL,
  scaling = 1000,
  norder = 3
)

Arguments

dat	input data following MDSINE's OTU table (i.e. variables in rows and samples in columns)
meta	metadata following MDSINE's metadata format
forceBreak	force to break the trajectory to handle pulsed perturbation (or species invasion) (default: NULL)
useSpline	use spline to smooth data and calculate gradients (default: TRUE)
dev	deviation (dev * mad) from the median to be considered as outliers (default: 5)
verbose	print more information (default: TRUE)
refSp	reference OTU for addative log ratio transformation (default: selected by BEEM)
warmup_iter	number of iterations to skip for estimating parameters (default: min_iter/2)
min_iter	minimal number of iterations for the EM algorithm (default: 30)
max_iter	maximal number of iterations for the EM algorithm (default: 100)
converge_thres	tolerance of change in mse to determine convergence (default: 1e-3)
ncpu	maximal number of CPUs used (default:2)
seed	seed used in BLASSO (default:NULL)
scaling	median total biomass to scale all biomass data (default:1000)
norder	order of spline basis (default: 3)

lch14forever/BEEM documentation built on April 5, 2025, 11:24 p.m.