RedoxRRI

Holobiont Redox Resilience Index for Integrative Stress Biology

---

Overview

RedoxRRI is an R package for computing and visualizing a Holobiont Redox Resilience Index (RRI) by integrating plant physiology, soil redox chemistry, and microbial resilience into a unified, directionally identifiable framework.

The package is designed for applications in redox ecology, plant–soil–microbe interactions, and holobiont resilience research, with explicit support for spatial structure, temporal dynamics, and the statistical challenges typical of ecological data.

---

Key Features

Multidomain integration

• Plant physiological stress buffering (e.g. ROS-related traits)
• Soil redox chemistry and electron-acceptor stability proxies
• Microbial resilience derived from abundance data and/or ecological networks

Directionally identifiable latent scores

• Dimension reduction via PCA, FA, NMF, WGCNA, and related approaches
• Explicit sign control with optional anchor variables to ensure reproducibility

Robustness to ecological data realities

• Median imputation within spatial strata
• Tolerance to MNAR missingness, zero inflation, and collinearity

Complementary outputs

• Absolute domain scores, scaled to ([0,1]), suitable for statistical modeling
• Compositional domain contributions, summing to 1, designed for ternary visualization

---

Conceptual Framework

Redox resilience at the holobiont scale emerges from interacting biological subsystems, rather than any single measurable trait.

RedoxRRI formalizes this concept by quantifying three domains:

• Physio — plant oxidative buffering and stress response
• Soil — redox chemistry and electron-acceptor stability
• Micro — microbial functional capacity and/or network resilience

Each domain is summarized using a latent variable derived from multivariate indicators. Domain scores are explicitly oriented so that higher values consistently indicate greater resilience, scaled to ([0,1]), and integrated into a composite Redox Resilience Index (RRI).

---

Design Principles

• Modular — domains can be added, removed, or reweighted
• Transparent — all transformations are inspectable and reproducible
• Flexible — supports multiple dimension-reduction strategies
• Mechanistic — designed for inference, not black-box prediction

---

Methodological Components

Dimension reduction

• Principal Component Analysis (PCA)
• Factor analysis
• Nonlinear embeddings (e.g. UMAP)
• Network-based summaries (e.g. WGCNA)

Microbial resilience metrics

• Abundance- or function-based representations
• Optional network topology metrics derived from ecological graphs

Aggregation

• User-defined weighting across biological domains
• Optional coupling terms to capture cross-domain coherence

---

Typical Workflow

1. Prepare domain-specific data

   • plant physiological traits
   • soil redox or chemical indicators
   • microbial abundance, functional profiles, or networks

2. Derive latent scores for each domain

3. Integrate domains into a composite Redox Resilience Index

4. Evaluate and compare

   • alternative weighting schemes
   • stress or disturbance scenarios
   • ecological or experimental contexts

---

Installation

install.packages("remotes")
remotes::install_github("mghotbi/RedoxRRI")

# with vignettes
install.packages("remotes")
install.packages("BiocManager")

BiocManager::install(c(
  "BiocStyle",
  "rmarkdown",
  "knitr"))

remotes::install_github(
  "mghotbi/RedoxRRI",
  build_vignettes = TRUE,
  dependencies = TRUE)

Example

library(RedoxRRI)

# Simulate a holobiont redox system
sim <- simulate_redox_holobiont(seed = 1)
str(sim)

# Compute the Redox Resilience Index
res <- rri_pipeline_st(
  ROS_flux     = sim$ROS_flux,
  Eh_stability = sim$Eh_stability,
  micro_data   = sim$micro_data,
  id           = sim$id,

  # Direction anchoring 
  direction_phys = "auto",
  direction_anchor_phys = "FvFm",
  direction_soil = "auto",
  direction_anchor_soil = "Eh"
)

# Absolute domain scores (for analysis)
head(res$row_scores)

# Compositional domain contributions (for ternary)
head(res$row_scores_comp)

# Ternary visualization
plot_RRI_ternary(res$row_scores_comp)

Conceptual overview of the RedoxRRI framework. Multivariate indicators from plant physiology, soil redox chemistry, and microbial systems are summarized into domain-level latent scores, directionally aligned, scaled, and integrated into a unified holobiont-scale Redox Resilience Index (RRI).

Output Structure

The primary output is an object of class RRI, containing: row_scores Absolute domain scores and per-sample RRI (all scaled to [0,1]) row_scores_comp Compositional domain contributions (Physio + Soil + Micro = 1) and RRI meta Metadata including model settings and the system-level RRI index (rri_index)

The compositional scores are directly suitable for ternary visualization.

Intended Use

RedoxRRI is designed for hypothesis-driven analysis of stress resilience in complex biological systems, with a focus on interpretability and mechanistic insight. The package is particularly well suited for: Ecological and environmental research Quantifying resilience across abiotic stress gradients Integrative stress biology Soil–plant–microbiome interaction studies Comparative and sensitivity analyses Method development and exploratory modeling

Design Philosophy

RedoxRRI is not a black-box predictive model. Instead, it prioritizes: explicit modeling choices traceable transformations from raw data to index values comparison of alternative hypotheses rather than automated optimization This makes RedoxRRI especially suitable for research contexts where understanding why resilience changes is as important as measuring how much.

Citation

If you use RedoxRRI in your research, please cite: Ghotbi, M. et al. RedoxRRI: A framework for holobiont-level redox resilience (manuscript in preparation)

License

GPL-3 © 2025 Mitra Ghotbi

Contact

Maintainer: Mitra Ghotbi
📧 mitra.ghotbi@gmail.com
🔗 ORCID: 0000-0001-9185-9993