paweh is statistical-genetics study-design software for prospective power and minimum sample size necessary (MSSN) calculations. Given a proposed sample size, it calculates power; given a target power, it calculates the sample size needed under the specified design assumptions. Its calculations are primarily parameter and model based and are intended for planning before data collection or genotyping, not for downstream association testing.
The package supports case-control association studies, affected-child trio designs using the Transmission Disequilibrium Test (TDT), and continuous, extreme-phenotype, and multivariate quantitative-trait designs. Depending on the design, calculations can account for genotype misclassification, phenotype misclassification, and locus heterogeneity. Canonical case-control and TDT interfaces support model-based and model-free workflows where those input modes are applicable.
Genetic study-design calculations are commonly formulated under homogeneous, error-free assumptions. Selected departures from those assumptions can materially affect power and MSSN while being difficult to incorporate analytically. paweh provides accessible, reproducible implementations of published methods for examining supported heterogeneity and misclassification mechanisms alongside conventional power and sample-size calculations.
paweh supports
For case-control NGS, paweh constructs a sequencing-to-called-genotype transition matrix and then applies the published Ahn / Chapman–Nam trend-test framework. TDT1-NGS implements the Kim (2015) raw-read likelihood and information framework under the documented fixed equal-depth multiplicative model. In the matching Kim AIS design, the package reproduces the reported requirements of 654 complete trios at 4x and 416 at 25x.
Model-based calculations express assumptions through quantities such as prevalence, disease-allele frequency, genotype relative risk, inheritance mode, or quantitative-trait variance. Where model-free input is supported, users can instead supply conditional genotype probabilities or expected transmission and non-transmission counts derived from pilot data, prior work, or external calculations. Model-free input does not eliminate assumptions; it places them in the supplied conditional quantities.
Most functions return structured objects containing both the primary design answer and intermediate quantities useful for validation. Examples below extract only the result needed for interpretation; consult the reference documentation when diagnostic components are required.
The complete documentation is available at https://akilanthony.github.io/paweh/.
Seven articles provide the main entry points:
Function-level reference documentation is also available from the pkgdown reference index.
paweh is currently under development and is not yet available through Bioconductor. Install the development version from GitHub:
if (!requireNamespace("remotes", quietly = TRUE)) {
install.packages("remotes")
}
remotes::install_github("akilanthony/paweh")paweh includes an interactive Shiny interface for:
The dashboard uses the same canonical package functions as the programmatic R interface and provides power, minimum-sample-size, sensitivity, visualization, advanced calculation details, and reproducible R calls. Construct and launch the dashboard explicitly with:
paweh_app() returns a Shiny application object and does not launch a browser or server on its own.
See the Interactive Dashboard vignette for a guided walkthrough of all three study-design workspaces.
The canonical case-control interface can calculate genotype-test power for a fixed design. This example assumes 500 cases, an equal number of controls, 5% disease prevalence, a disease-allele frequency of 0.30, and a genotype relative risk of 1.8 under the package’s multiplicative model:
library(paweh)
power_result <- cc_power(
N_case = 500,
alpha = 0.05,
input_mode = "model_based",
prev = 0.05,
pd = 0.30,
R2 = 1.8,
MOI = "M",
k = 1,
verbose = FALSE
)
power_result$tests$genotypes$power
#> [1] 0.8366557The inverse calculation asks how many subjects are required for 80% power under the same assumptions:
mssn_result <- cc_mssn(
power = 0.80,
alpha = 0.05,
input_mode = "model_based",
prev = 0.05,
pd = 0.30,
R2 = 1.8,
MOI = "M",
k = 1,
verbose = FALSE
)
c(
cases = mssn_result$tests$genotypes$MSSN_case,
controls = mssn_result$tests$genotypes$MSSN_ctrl,
total = mssn_result$tests$genotypes$MSSN_total
)
#> cases controls total
#> 457 457 914These are prospective operating characteristics under the stated model, not estimates fitted from subject-level genomic data. See the case-control article for model-free inputs and supported error and heterogeneity scenarios.
Power and MSSN are inverse design questions, but integer recruitment targets may not reproduce a requested power exactly because required sample sizes are rounded up. Results also inherit all stated distributional, prevalence, frequency, relative-risk, linkage-disequilibrium, and error assumptions. The study-design articles document these assumptions and identify settings that are outside the current implementation.
| Design | Primary functions | Documentation |
|---|---|---|
| Case-control association |
cc_power(), cc_mssn()
|
Case-Control Study Design |
| Affected-child trios / TDT |
tdt_power(), tdt_mssn()
|
TDT Study Design |
| Continuous quantitative trait |
qtl_anova_power(), qtl_anova_mssn()
|
Quantitative-Trait Study Design |
| Extreme quantitative trait |
qtl_threshold_chisq_power(), qtl_threshold_chisq_mssn()
|
Quantitative-Trait Study Design |
| Multivariate quantitative traits |
qtl_multivariate_power_full(), qtl_multivariate_mssn_full()
|
Quantitative-Trait Study Design |
| Case-control NGS |
cc_ngs_power(), cc_ngs_mssn()
|
Sequencing-Based Genetic Study Design |
| TDT1-NGS |
tdt_ngs_power(), tdt_ngs_mssn()
|
Sequencing-Based Genetic Study Design |
| Error-aware case-control / LRTae |
lrtae_ncp(), lrtae_power(), lrtae_mssn()
|
LRTae Power and Sample-Size Design |
paweh includes literature-backed regression tests and worked examples for case-control genotype and phenotype misclassification, TDT phenotype misclassification and locus heterogeneity, and multivariate quantitative-trait power and sample size. The articles distinguish exact or near-exact published-example reproductions from papers used only as methodological context or motivating applications.
paweh is a prospective statistical study-design package. It does not perform GWAS association testing, sequence processing, variant annotation, or general genomic data management. Instead, it is intended to complement downstream genetic-analysis workflows by helping researchers examine power, sample size, and sensitivity to supported design assumptions before collecting data or performing genotyping.