Skip to contents

Simulate a Holobiont Redox Dataset with Spatio-Temporal Structure

Source: R/simulate_redox_holobiont.R
simulate_redox_holobiont.Rd

Generate a synthetic plant–soil–microbiome dataset with (i) a latent redox state evolving over time, (ii) a mid-season disturbance pulse, (iii) partial cross-domain decoupling, (iv) configurable microbial sparsity (zero inflation), and (v) missing-not-at-random (MNAR) Eh dropout under strongly reduced states.

The simulator is designed for benchmarking, validation, and demonstration of RedoxRRI workflows. It is not intended to represent any real ecosystem.

Usage

simulate_redox_holobiont(
  n_plot = 4,
  n_depth = 2,
  n_plant = 5,
  n_time = 15,
  p_micro = 60,
  seed = 123,
  include_graph = FALSE,
  depth_labels = NULL,
  disturbance_strength = 0.4,
  disturbance_center = NULL,
  disturbance_width = 0.1,
  seasonal_amp = 0.08,
  seasonal_phase = 0,
  stochastic_reassembly = TRUE,
  decoupling = 0.3,
  zero_inflation = 0.4,
  MNAR_strength = 0.4,
  Eh_dropout_threshold = 0.25,
  micro_mean = 8,
  micro_slope = 3,
  micro_lambda_min = 1e-08,
  micro_lambda_max = 1e+06
)

Arguments

n_plot

Integer >= 1. Number of plots.

n_depth

Integer >= 1. Number of soil depth strata.

n_plant

Integer >= 1. Number of plants per plot–depth.

n_time

Integer >= 2. Number of time points per plant.

p_micro

Integer >= 1. Number of microbial features.

seed

Optional integer seed. Use NULL for stochastic runs.

include_graph

Logical. If TRUE and igraph is available, return a simulated network as graph.

depth_labels

Optional character vector of length n_depth. If NULL, depths are labeled "D1", "D2", ....

disturbance_strength

Numeric in [0, 1]. Magnitude of disturbance pulse.

disturbance_center

Optional numeric. If NULL, centered at median time.

disturbance_width

Numeric > 0. Pulse width as fraction of season length.

seasonal_amp

Numeric >= 0. Amplitude of seasonal forcing.

seasonal_phase

Numeric. Phase shift for seasonal sine wave (radians).

stochastic_reassembly

Logical. If TRUE, microbial counts include a stochastic reassembly component (mixture of intensities).

decoupling

Numeric in [0, 1]. Cross-domain decoupling (0=tightly coupled).

zero_inflation

Numeric in [0, 1]. Probability a microbial entry is set to 0.

MNAR_strength

Numeric in [0, 1]. Strength of MNAR Eh dropout in reduced states.

Eh_dropout_threshold

Numeric in [0, 1]. Latent threshold below which dropout can occur.

micro_mean

Numeric > 0. Baseline mean intensity for microbial counts.

micro_slope

Numeric >= 0. Coupling strength between latent redox state and microbial intensity.

micro_lambda_min

Numeric > 0. Lower bound for microbial Poisson intensity.

micro_lambda_max

Numeric > 0. Upper bound for microbial Poisson intensity.

Value

A named list with:

  • id: Data frame of experimental design (plot, depth, plant_id, time).

  • ROS_flux: Plant physiological indicators (data frame).

  • Eh_stability: Soil redox chemistry variables (data frame; Eh may include NA).

  • micro_data: Microbial feature matrix (data frame; non-negative integers).

  • latent_truth: Underlying latent redox state in [0, 1].

  • graph: Optional igraph object if requested and available.

Examples

# ---- Simulate small holobiont dataset ----
sim <- simulate_redox_holobiont(
  n_plot = 2,
  n_depth = 1,
  n_plant = 2,
  n_time = 8,
  p_micro = 20,
  seed = 1
)

# ---- Compute RedoxRRI ----
res <- rri_pipeline_st(
  ROS_flux = sim$ROS_flux,
  Eh_stability = sim$Eh_stability,
  micro_data = sim$micro_data,
  id = sim$id,
  reducer = "per_domain",
  scaling = "pnorm"
)

# ---- Correlation with known latent truth (simulation only) ----
cor_val <- rri_latent_correlation(
  res,
  latent_truth = sim$latent_truth,
  method = "pearson"
)