Skip to contents

ordinalMIC

Minimum Inhibitory Concentration (MIC*) Estimation and groupwise comparisons from proportional-odds ordinal regression models.

Package website: https://clstacy.github.io/ordinalMIC/

Companion Website: https://LewisLabUARK.github.io/MICalculator/

DOI

Purpose

Microbiological assays often record growth on an ordered score scale (e.g., 0–4) across concentrations. Treating these scores as binary or continuous can lead to inaccurate conclusions. ordinalMIC turns cumulative-link models (ordinal::clm) into interpretable concentration-scale summaries:

  • MIC* per group (the concentration where predicted probability of “no growth” is 50%)
  • Δ-MIC* (additive differences)
  • MIC* ratio (fold-change)
  • Difference-of-differences (DoD) on additive and ratio scales to test interactions

It also provides compact output tables, a ggplot2::autoplot() for quick figures generation.

Installation

Stable (from r-universe) 

install.packages("ordinalMIC", repos = c("https://clstacy.r-universe.dev"))

From source tarball 

install.packages("ordinalMIC_1.0.0.tar.gz", repos = NULL, type = "source")

Dependencies

  • Required: ggplot2, stats, utils, cli, tibble, R (>= 4.1)
  • Suggested: ordinal for modeling

Core Idea

Fit a proportional-odds ordinal regression to your scored outcomes across concentration. Then use mic_solve() to:

  1. Compute group-specific MIC* on the original concentration scale (using your chosen transform/inverse).
  2. Compare groups by ΔMIC* and MIC ratios with CIs and p-values.
  3. Test interactions via difference-of-differences (additive and ratio scales).

Default transform is log1p for concentration; override if your model uses a different link-side transform.

Quick Start 

library(ordinalMIC)

# Example data included with the package
data(yeast_df)

# Fit a proportional-odds model (requires 'ordinal' package)
fit <- ordinal::clm(score ~ strain * treatment + log1p(conc), data = yeast_df)

# Estimate MIC* and compare groups
res <- mic_solve(
  clm_fit    = fit,
  conc_name  = "conc",
  transform_fun     = log1p,
  inv_transform_fun = expm1,
  alpha      = 0.05,
  compare_pairs = "all"
)

# Tables
head(tidy.mic_solve(res, table = "mic"))
head(tidy.mic_solve(res, table = "ratio"))

# Plots
ggplot2::autoplot(res, type = "mic")

Main functions

mic_solve()

  • Inputs: fitted ordinal::clm, newdata, conc_name, transform/inverse functions
  • Outputs: MICs, Δ-MIC, ratios, DoD summaries, all with CIs and p-values

ggplot2::autoplot.mic_solve()

  • Quick visualization of MICs, Δ-MICs, ratios, DoDs

tidy.mic_solve()

  • Extracts summary tables for MICs, Δ-MICs, ratios, DoDs as tibble objects

Data

Example dataset yeast_df included with the package:

Column Type Description
strain factor Yeast strain (WT, Mut)
treatment factor Treatment type (None, Salt)
conc numeric Concentration (e.g., drug concentration)
score ordered Growth score (0 = no growth, 4 = full growth)
rep factor Replicate identifier

Modeling guidance

  • Outcome must be an ordered factor with levels increasing in severity
  • Transform concentration (default log1p); supply both transform and inverse
  • newdata gives groups to estimate MIC*, automatically generated if not given.
  • Use DoD (Difference of Difference) summaries to interpret interaction effects on MIC*

MIC definition

MIC is the concentration where the predicted probability of “no growth” equals 0.5, computed from the fitted clm.

  • MIC ratios: differences in predicted log-MIC

  • DoD ratios: ratio-of-ratios (interaction on log scale)

Troubleshooting

  • Outcome not ordered: convert with ordered()
  • Transform mismatch: supply correct transform/inverse
  • No Estimate for MIC: reconsider dose range (censored results/no change observed)
  • Comparisons: restrict with compare_pairs = "share_any"

Citation

If you use ordinalMIC in published research, please cite: Stacy C (2025). ordinalMIC: Minimum Inhibitory Concentration Estimation from Ordinal Regression Models. R package version 1.0.0, https://clstacy.github.io/ordinalMIC/.

Contributing

License

MIT License. See LICENSE file for details.