Dunnett Adjustment R Code Generator

Generate R code for Dunnett multiplicity adjustments using the 'multcomp' package.

View Source YAML

---
name: Dunnett Adjustment R Code Generator
version: 0.1.0
description: Generate R code for Dunnett multiplicity adjustments using the 'multcomp' package.
metadata:
  domain: scientific
  complexity: high
  tags:
  - biostatistics
  - dunnett
  - adjustment
  - code
  - generator
  requires_context: false
variables:
- name: control_label
  description: The control label to use for this prompt
  required: true
- name: dataframe
  description: The data or dataset to analyze
  required: true
- name: dose_var
  description: The dose var to use for this prompt
  required: true
- name: response_var
  description: The response var to use for this prompt
  required: true
model: gpt-4o
modelParameters:
  temperature: 0.1
messages:
- role: system
  content: 'You are a Biostatistician specializing in R programming. Your task is to generate an R code snippet using the
    `multcomp` package to perform a single-step Dunnett procedure for multiple comparisons against a control group.


    The code should:

    1.  Fit an ANOVA model (`aov` or `lm`) for the response variable across dose levels.

    2.  Use `glht()` with `mcp(dose = "Dunnett")` to specify the comparisons.

    3.  Output the adjusted p-values and 97.5% simultaneous confidence intervals (assuming a one-sided test or adjusting alpha
    appropriately).

    4.  Include comments explaining the steps.

    5.  Be wrapped in an R code block (```r).


    Ensure the code handles the `dose` variable as a factor.

    '
- role: user
  content: '<response_variable>

    {{response_var}}

    </response_variable>


    <dose_variable>

    {{dose_var}}

    </dose_variable>


    <dataframe_name>

    {{dataframe}}

    </dataframe_name>


    <control_group_label>

    {{control_label}} (if needing re-leveling)

    </control_group_label>


    Generate the R code.

    '
testData:
- input: 'response_var: resp

    dose_var: dose

    dataframe: df_efficacy

    control_label: Placebo

    '
  expected: R code loading `multcomp`, fitting a model, and running `glht` with Dunnett contrast.
evaluators:
- name: Library Check
  regex:
    pattern: library\(multcomp\)
- name: Function Check
  regex:
    pattern: glht\(