Package 'r4pde'

Description

This function performs the analysis of a simple method introduced by Nelson (1996) and expanded by Laranjeira et al. (1998). The function assumes the dataframe supplied as input has columns 'x', 'y', and 'i', where 'x' and 'y' are spatial coordinates and 'i' is a disease indicator variable (1 if diseased, otherwise 0). The function performs several steps including filtering rows where 'i' is 1, converting to an adjacency matrix, and creating foci using igraph. It then calculates various statistics about the foci and returns these in a list.

Usage

AFSD(df)

Arguments

Value

A list containing: cluster_summary2: a dataframe summarizing the number and size of foci, and proportions of diseased plants. cluster_df: a dataframe containing foci information, including size and number of rows and columns in each foci. df_clustered: the original dataframe with an added 'focus_id' column, showing which foci each row belongs to.

See Also

Other Spatial analysis: BPL(), count_subareas(), count_subareas_random(), fit_gradients(), join_count(), oruns_test(), oruns_test_boustrophedon(), oruns_test_byrowcol(), plot_AFSD()

Examples

# Generate a sample dataframe
set.seed(123)
df <- data.frame(x = sample(1:100, 500, replace = TRUE),
                 y = sample(1:100, 500, replace = TRUE),
                 i = sample(0:1, 500, replace = TRUE, prob = c(0.7, 0.3)))

# Perform the AFSD
result <- AFSD(df)

Description

Augment functional PCA

Usage

augment_functional_pca(x)

Arguments

Description

Wheat blast dataset with severity and weather covariates.

Usage

BlastWheat

Format

A data frame with the following columns:

heading

Date of heading

inc_mean

Mean incidence

index_mean

FHB index mean

latitude

Latitude coordinate

location

Experimental site name

longitude

Longitude coordinate

state

Brazilian state

study

Study ID or code

year

Crop year

yld_mean

Mean yield

Source

Del Ponte Lab internal data

Description

This function calculates the Binary Power Law (BPL) parameters for spatial disease patterns, fits a linear model, and performs a hypothesis test for the slope.

Usage

BPL(data)

Arguments

Details

The function performs the following steps:

1. Summarizes the data by field to calculate the total number of observations (n_total), mean incidence (incidence_mean), observed variance (V), and binomial variance (Vbin).

2. Log-transforms the variances.

3. Fits a linear model to the log-transformed variances.

4. Tests the hypothesis that the slope of the linear model is equal to 1.

Value

A list containing the following elements:

• summary: A data frame summarizing the input data by field, including total observations (n_total), mean incidence (incidence_mean), observed variance (V), and binomial variance (Vbin).

• model_summary: A summary of the linear model fitted to the log-transformed variances.

• hypothesis_test: The result of the hypothesis test for the slope being equal to 1.

• ln_Ap: The intercept of the linear model, representing the natural logarithm of the parameter \( A_p \).

• slope: The slope of the linear model.

See Also

Other Spatial analysis: AFSD(), count_subareas(), count_subareas_random(), fit_gradients(), join_count(), oruns_test(), oruns_test_boustrophedon(), oruns_test_byrowcol(), plot_AFSD()

Examples

# Example usage with a sample data frame
result <- BPL(FHBWheat)
print(result$summary)
print(result$model_summary)
print(result$hypothesis_test)
print(paste("ln(Ap):", result$ln_Ap))
print(paste("Slope (b):", result$slope))

Description

Soybean bud blight incidence in experimental blocks.

Usage

BudBlightSoybean

Format

A data frame with the following columns:

block

Block number

time

Time point of assessment

treat

Treatment name

y

Incidence or severity value

Source

Del Ponte Lab internal data

Description

Fits a generalized additive model (GAM) to replicated disease progress curves. By default, the response is modeled with beta regression (logit link) after a Smithson–Verkuilen adjustment to handle 0 and 1 values; if beta precision (theta) estimation fails, the model falls back to a quasibinomial GAM fit on the unclipped proportion response.

From the fitted model, the function derives (i) environment-adjusted treatment mean curves on a common time grid, (ii) L2 functional distances among treatment mean trajectories followed by hierarchical clustering, and (iii) curve-level fitted trajectories (excluding unit random effects) to build a curve-by-curve distance matrix. Optionally, it performs a restricted permutation test on the curve-level distance matrix for a priori group labels, and computes bootstrap envelopes and distance uncertainty summaries from resampled predicted curves.

Usage

compare_curves(
  data,
  time,
  response,
  treatment,
  environment = NULL,
  block = NULL,
  unit = NULL,
  response_scale = c("proportion", "percent"),
  eps = 1e-04,
  min_points = 5,
  grid_n = 140,
  env_ref = NULL,
  k_smooth = 10,
  k_env = 4,
  k_trt = 6,
  gamma = 1.4,
  discrete = TRUE,
  family_try = c("betar", "quasibinomial"),
  cluster_k = 4,
  hc_method = "ward.D2",
  test_factor = NULL,
  n_perm = 999,
  test_mode = c("auto", "none", "global", "global_pairwise"),
  min_strata_for_pairwise = 6,
  perm_unit = NULL,
  perm_strata = NULL,
  bootstrap = FALSE,
  boot_B = 399,
  boot_seed = 1,
  boot_ci = c(0.025, 0.975),
  bootstrap_mode = c("predicted", "refit"),
  show_progress = TRUE,
  curve_level = NULL,
  ...
)

Arguments

Details

Functional distances among treatments are computed as an L2 norm over the predicted mean curves evaluated on a common time grid. Curve-level distances are computed analogously using fitted trajectories (excluding unit random effects), and are scaled by the median non-zero distance for numerical stability; interpret these distances as relative rather than absolute units.

The permutation test uses a PERMANOVA-like pseudo-$F$ statistic computed from the curve-level distance matrix, with optional restricted permutations within strata. When enabled, pairwise comparisons are adjusted using Holm's method.

Value

An object of class "r4pde_compare_curves" containing:

• gam: fitted GAM object and family_used;

• pred: environment-adjusted treatment mean curves on the common grid;

• distance: treatment-by-treatment functional distance matrix and hc;

• clusters: treatment cluster assignments and summary scores;

• curve_distance: curve-by-curve distance matrix;

• test: optional distance-based permutation test results;

• bootstrap: optional bootstrap envelopes and distance summaries;

• diagnostic fields including settings and captured beta-fit warnings.

See Also

plot_curves, plot_dendrogram, diagnose_curves

Description

Survival analysis for quantitative ordinal scale data

Usage

CompMuCens(dat, scale, grade = TRUE, ckData = FALSE)

Arguments

Details

This function supports analysis of quantitative ordinal scale data via interval-censored methods. The approach follows the workflow described by Chiang et al. (2023) and the reference implementation provided in the CompMuCens repository.

Value

A list containing the score statistic, hypothesis tests, adjusted significance level, and conclusion based on pairwise comparisons.

References

Chiang, K.S., Chang, Y.M., Liu, H.I., Lee, J.Y., El Jarroudi, M. and Bock, C. (2023). Survival Analysis as a Basis to Test Hypotheses When Using Quantitative Ordinal Scale Disease Severity Data. Phytopathology. https://apsjournals.apsnet.org/doi/abs/10.1094/PHYTO-02-23-0055-R

See Also

Other Disease quantification: DSI(), DSI2()

Examples

if (requireNamespace("interval", quietly = TRUE)) {
  trAs <- c(5,4,2,5,5,4,4,2,5,2,2,3,4,3,2,2,6,2,2,4,2,4,2,4,5,3,4,2,2,3)
  trBs <- c(5,3,2,4,4,5,4,5,4,4,6,4,5,5,5,2,6,2,3,5,2,6,4,3,2,5,3,5,4,5)
  trCs <- c(2,3,1,4,1,1,4,1,1,3,2,1,4,1,1,2,5,2,1,3,1,4,2,2,2,4,2,3,2,2)
  trDs <- c(5,5,4,5,5,6,6,4,6,4,3,5,5,6,4,6,5,6,5,4,5,5,5,3,5,6,5,5,5,6)
  inputData <- data.frame(
    treatment = c(rep("A",30), rep("B",30), rep("C",30), rep("D",30)),
    x = c(trAs, trBs, trCs, trDs)
  )
  CompMuCens(dat = inputData,
             scale = c(0,3,6,12,25,50,75,88,94,97,100,100),
             ckData = TRUE)
}

Description

This function takes a binary matrix (0s and 1s) and divides it into rectangular subareas, counting the number of ones in each. Subareas are defined by the number of rows and columns specified by the user. If the matrix dimensions are not perfectly divisible by the subarea size, edge subareas may be smaller.

Usage

count_subareas(matrix_data, sub_rows, sub_cols)

Arguments

Value

A matrix where each cell corresponds to a subarea and contains the count of ones.

See Also

Other Spatial analysis: AFSD(), BPL(), count_subareas_random(), fit_gradients(), join_count(), oruns_test(), oruns_test_boustrophedon(), oruns_test_byrowcol(), plot_AFSD()

Examples

set.seed(123)
mat <- matrix(sample(c(0, 1), 12 * 16, replace = TRUE), nrow = 16, ncol = 12)
count_matrix <- count_subareas(mat, sub_rows = 3, sub_cols = 3)
print(count_matrix)

Description

Randomly samples submatrices (quadrats) of specified size from a binary matrix, and returns the positions, submatrices, and count of 1s in each sampled quadrat.

Usage

count_subareas_random(matrix_data, sub_rows = 3, sub_cols = 3, n_samples = 100)

Arguments

Value

A list of sampled subgrids. Each element is a list with:

See Also

Other Spatial analysis: AFSD(), BPL(), count_subareas(), fit_gradients(), join_count(), oruns_test(), oruns_test_boustrophedon(), oruns_test_byrowcol(), plot_AFSD()

Description

Computes residuals, fitted values, and model diagnostics for GAM-based epidemic curve models fitted with compare_curves(). Produces diagnostic plots without invoking base graphics.

Usage

diagnose_curves(x, grid_n = 200)

Arguments

Value

An object of class "r4pde_curve_diagnostics" containing:

• diagnostic data,

• k-index checks,

• concurvity estimates,

• ggplot-based diagnostic panels.

Description

Assessment of Didymella symptoms in watermelon plots.

Usage

DidymellaWatermelon

Format

A data frame with:

EW_row

Row position (east–west)

NS_col

Column position (north–south)

dap

Days after planting

severity

Disease severity

Source

Del Ponte Lab internal data

Description

This function calculates the Disease Severity Index (DSI) based on the provided unit, class, and maximum class value. The DSI is computed by aggregating the classes, calculating weights by multiplying the frequency of each class by the class itself, and then dividing the sum of these weights by the product of the total number of entries and the maximum class value, then multiplying by 100.

Usage

DSI(unit, class, max)

Arguments

Value

Returns a single numeric value representing the DSI.

See Also

Other Disease quantification: CompMuCens(), DSI2()

Examples

# Example usage:
unit <- c(1, 2, 3, 4, 5, 6)
class <- c(1, 2, 1, 2, 3, 1)
max <- 3
DSI(unit, class, max)

Description

This function calculates the Disease Severity Index (DSI) given a vector of classes, a vector of frequencies, and a maximum possible class value. The DSI is calculated as a weighted sum of class values, where each class is multiplied by its corresponding frequency, then divided by the product of the total frequency and maximum class value, and finally multiplied by 100 to get a percentage.

Usage

DSI2(class, freq, max)

Arguments

Value

Returns a single numeric value representing the DSI.

See Also

Other Disease quantification: CompMuCens(), DSI()

Examples

DSI2(c(0, 1, 2, 3, 4), c(2, 0, 5, 0, 5), 4)

Description

Fusarium head blight quadrat assessments in wheat.

Usage

FHBWheat

Format

A data frame with:

field

Field identifier

i

Row position

n

Column position

quadrat

Quadrat ID

season

Crop season

Source

Del Ponte Lab internal data

Description

This function fits three gradient models (exponential, power, and modified power) to given data. It then ranks the models based on their R-squared values and returns diagnostic plots for each model.

Usage

fit_gradients(data, C = 1)

Arguments

Value

A list containing:

See Also

Other Spatial analysis: AFSD(), BPL(), count_subareas(), count_subareas_random(), join_count(), oruns_test(), oruns_test_boustrophedon(), oruns_test_byrowcol(), plot_AFSD()

Examples

x <- c(0.8, 1.6, 2.4, 3.2, 4, 7.2, 12, 15.2, 21.6, 28.8)
Y <- c(184.9, 113.3, 113.3, 64.1, 25, 8, 4.3, 2.5, 1, 0.8)
grad1 <- data.frame(x = x, Y = Y)
library(ggplot2)
mg <- fit_gradients(grad1, C = 0.4)
mg$plot_power_original +
  labs(title = "", x = "Distance from focus (m)", y = "Count of lesions")

Description

Fits a generalized additive model (GAM) to replicated disease progress curves, returning adjusted treatment mean curves evaluated on a common time grid.

Usage

functional_curves(
  data,
  time,
  response,
  treatment,
  environment = NULL,
  block = NULL,
  unit = NULL,
  response_scale = c("proportion", "percent"),
  eps = 1e-04,
  min_points = 5,
  grid_n = 140,
  env_ref = NULL,
  k_smooth = 10,
  k_env = 4,
  k_trt = 6,
  gamma = 1.4,
  discrete = TRUE,
  family_try = c("betar", "quasibinomial"),
  show_progress = TRUE,
  covariates = NULL,
  include_covariates = FALSE,
  covariate_smooths = FALSE,
  ...
)

Arguments

Details

Genotype-level covariates are descriptors that do not vary within a genotype, such as phenological groups. For example, in wheat blast studies, a cultivar's heading_group (early, intermediate, late) can be supplied. This helps distinguish between true genetic resistance and phenological escape, as cultivars with different heading dates may encounter different infection-risk windows in the same environment. When include_covariates = TRUE, the model accounts for these covariates, yielding heading-adjusted functional resistance.

Value

An object of class "functional_curves" containing:

• gam: fitted GAM object and family_used;

• curves: environment-adjusted treatment mean curves on the common grid;

• grid: the time grid used;

• observed_data: the processed input data;

• genotype_info: a tibble with one row per genotype containing covariates;

• vars: variable names used;

• settings: model settings;

• warnings_betar: any warnings caught during beta fitting;

• plot_mean: ggplot object of the mean curves.

Examples

## Not run: 
fc <- functional_curves(
  data = my_data,
  time = "time_var",
  response = "severity",
  treatment = "cultivar",
  environment = "env",
  covariates = c("heading_group"),
  include_covariates = TRUE,
  covariate_smooths = TRUE
)
plot(fc)

## End(Not run)

Description

Computes L2 functional distances among treatment mean trajectories evaluated on a common time grid from a functional_curves object. Also computes curve-level distances, performs hierarchical clustering, and optionally performs permutation testing.

Usage

functional_distances(
  object,
  cluster_k = 4,
  hc_method = "ward.D2",
  test_factor = NULL,
  n_perm = 999,
  test_mode = c("auto", "none", "global", "global_pairwise"),
  min_strata_for_pairwise = 6,
  perm_unit = NULL,
  perm_strata = NULL,
  bootstrap = FALSE,
  boot_B = 399,
  boot_seed = 1,
  boot_ci = c(0.025, 0.975),
  show_progress = TRUE,
  curve_level = NULL,
  ...
)

Arguments

Value

An object of class "functional_distances" containing:

• distance: treatment-by-treatment functional distance matrix;

• distance_table: pairwise distances as a data frame;

• hc: hclust object;

• clusters: treatment cluster assignments;

• curve_distance: curve-by-curve distance matrix;

• test: permutation test results;

• bootstrap: bootstrap results if requested.

Examples

## Not run: 
fd <- functional_distances(fc, cluster_k = 4)
print(fd)
plot_dendrogram(fd)
plot_curves(fd)

## End(Not run)

Description

Computes normalized functional instability (NFI) for each treatment based on genotype-by-environment predicted curves extracted from a compare_curves() object. Optionally, instability can be decomposed into spatial and temporal components if the environment identifier can be split into location and year.

Usage

functional_instability(
  x,
  n_time = 200,
  env_sep = NULL,
  env_names = c("location", "year"),
  return_curves = FALSE
)

Arguments

Details

The overall instability metric is defined as the mean integrated squared deviation of each genotype-by-environment curve from the genotype-specific mean curve across environments, normalized by the integrated squared mean curve.

Let $f_{ge}(t)$ denote the predicted epidemic trajectory of genotype $g$ in environment $e$, and let $\bar f_g(t)$ denote the mean trajectory of genotype $g$ across environments. Functional instability is computed as:

$FI_g = \frac{1}{E_g}\sum_{e=1}^{E_g} \int_T \left(f_{ge}(t)-\bar f_g(t)\right)^2 dt$

and normalized as:

$nFI_g = \frac{FI_g}{\int_T \bar f_g(t)^2 dt}$

Numerical integration is performed with the trapezoidal rule on a regular prediction grid over the observed time domain.

Value

If return_curves = FALSE, a tibble with one row per genotype and columns:

geno

Treatment or genotype identifier.

n_env

Number of environments used in the calculation.

FI

Absolute functional instability.

mean_energy

Integrated squared mean epidemic curve.

nFI

Normalized functional instability. Lower values indicate greater stability.

If env_sep is provided, the returned tibble also includes:

FI_space, mean_energy_space, nFI_space

Spatial component of instability.

FI_time, mean_energy_time, nFI_time

Temporal component of instability.

If return_curves = TRUE, a list is returned with two elements:

metrics

The tibble described above.

curves

A tibble of predicted genotype-by-environment curves with columns geno, env, time, mu, and eta.

See Also

functional_curves, compare_curves

Examples

## Not run: 
m1 <- r4pde::compare_curves(
  data = dat_ready,
  time = "time",
  response = "y",
  treatment = "geno",
  environment = "env",
  cluster_k = 4
)

# Overall instability
res <- functional_instability(m1)
res

# Overall + spatial and temporal components
# Assuming env has values such as "PF_2021"
res2 <- functional_instability(m1, env_sep = "_")
res2

# Return curves used in the calculations
out <- functional_instability(m1, env_sep = "_", return_curves = TRUE)
out$metrics
head(out$curves)

## End(Not run)

Description

Performs functional principal component analysis on fitted disease progress curves returned by functional_curves. The function decomposes variation among epidemic trajectories into orthogonal temporal components and returns curve-level scores, eigenfunctions, variance explained, and reconstructed curves.

Usage

functional_pca(
  object,
  n_components = NULL,
  var_explained = 0.95,
  center = TRUE,
  scale = FALSE,
  method = c("pca_on_grid"),
  ...
)

Arguments

Details

The function uses the fitted curves from functional_curves() and does not refit the disease progress model. The first implementation uses PCA on a common prediction grid. FPC scores can be analyzed as functional epidemiological traits in downstream models.

Interpretation:

• FPC1 often captures the largest mode of variation, commonly overall epidemic intensity or speed.

• Later FPCs may capture timing of disease onset, curve crossing, late acceleration, or other shape-related deviations.

• Interpretation must be data-driven and should be based on eigenfunction plots and mean ± perturbation plots.

Value

An object of class "r4pde_functional_pca" containing:

• scores: Tibble of curve-level FPC scores.

• eigenfunctions: Tibble of eigenfunction values across the time grid.

• variance: Tibble of eigenvalues, variance explained, and cumulative variance.

• mean_curve: Tibble of the mean curve.

• reconstructed: Tibble of reconstructed curves using retained components.

• input_curves: Tibble of original fitted curves.

• pca: The underlying prcomp object.

• settings: List of settings used.

• call: The matched call.

Examples

## Not run: 
curves <- functional_curves(...)
fpca <- functional_pca(curves, var_explained = 0.95)

print(fpca)
plot(fpca, type = "scree")
plot(fpca, type = "components")
plot(fpca, type = "scores", components = c(1, 2))

scores <- get_fpca_scores(fpca)

## End(Not run)

Description

Cluster curves based on FPCA scores

Usage

functional_pca_clusters(
  x,
  k = NULL,
  components = NULL,
  method = c("kmeans", "hclust"),
  choose_k = c("none", "silhouette", "elbow"),
  ...
)

Arguments

Description

Computes a functional resistance index (FRI) relative to a reference genotype curve. Optionally adjusts FRI by a functional instability penalty to calculate a stability-adjusted functional resistance index (SAFRI). Supports grouping genotypes into classes based on quantiles, clustering, or bootstrap-supported differences.

Usage

functional_resistance(
  object,
  instability = NULL,
  reference,
  lambda = 1,
  method = c("positive_area", "l2_difference"),
  scale_nfi = FALSE,
  n_groups = 4,
  group_method = c("none", "quantile", "bootstrap", "clustering"),
  time_var = NULL,
  genotype_var = NULL,
  n_boot = 1000,
  ci_level = 0.95,
  clustering_method = c("hclust", "kmeans"),
  adjust_by = NULL,
  reference_within_group = FALSE,
  group_within_adjust_by = TRUE,
  ...
)

Arguments

Value

A list with a table containing genotype, FRI, nFI, SAFRI, rank, and classes, and optionally bootstrap results.

Examples

## Not run: 
fr <- functional_resistance(
  fc,
  reference = "Susceptible",
  group_method = "bootstrap"
)
print(fr)

## End(Not run)

Description

Observations of Fusarium symptoms in banana fields.

Usage

FusariumBanana

Format

A data frame with:

field

Field ID

lat

Latitude

lon

Longitude

marker

Infection marker presence

Source

Del Ponte Lab internal data

Description

This function extracts daily weather data from the Brazilian Daily Weather Gridded Data (BR-DWGD) NetCDF files for specified locations and dates. It mimics the functionality of get_nasapower and get_era5 by returning a daily-aggregated data frame with columns named according to nasapower conventions.

Usage

get_brdwgd(
  data,
  days_around,
  date_col,
  study_col = "study",
  path = "netcdf_files",
  vars = c("pr", "Tmax", "Tmin", "Rs", "RH"),
  direction = "both"
)

Arguments

Details

The function requires the terra package to read NetCDF files and the tidyr package for data reshaping. It expects NetCDF files to be named starting with the variable name (e.g., pr_*.nc) and to contain the variable with the same name.

Value

A data frame (tibble) with daily weather data. Columns are named to match nasapower conventions where possible: date, PRECTOTCORR (from pr), T2M_MAX (from Tmax), T2M_MIN (from Tmin), ALLSKY_SFC_SW_DWN (from Rs), RH2M (from RH), longitude, latitude, and study.

See Also

Other Disease modeling: get_era5(), get_nasapower(), windowpane()

Description

This function downloads ERA5 historical weather data from the Open-Meteo API for specified locations and dates. It mimics the functionality of get_nasapower by fetching weather variables and returning a daily-aggregated data frame. It uses the Open-Meteo Archive API and aggregates hourly data to daily values to ensure all variables (including relative humidity and dew point) are available.

Usage

get_era5(
  data,
  days_around,
  date_col,
  study_col = "study",
  pars = c("temperature_2m", "relative_humidity_2m", "precipitation", "dewpoint_2m"),
  models = NULL,
  direction = "both"
)

Arguments

Details

Since the Open-Meteo Archive API daily endpoint does not provide all variables (like mean relative humidity) directly, this function fetches hourly data and performs the daily aggregation manually:

• T2M: Mean of hourly temperature_2m

• T2M_MAX: Max of hourly temperature_2m

• T2M_MIN: Min of hourly temperature_2m

• RH2M: Mean of hourly relative_humidity_2m

• PRECTOTCORR: Sum of hourly precipitation

• T2MDEW: Mean of hourly dewpoint_2m

Value

A data frame (tibble) with daily weather data. Columns are named to match nasapower conventions: date, T2M, T2M_MAX, T2M_MIN, RH2M, PRECTOTCORR, T2MDEW, longitude, latitude, and study.

See Also

Other Disease modeling: get_brdwgd(), get_nasapower(), windowpane()

Description

Get FPCA eigenfunctions

Usage

get_fpca_eigenfunctions(x, components = NULL)

Arguments

Description

Get FPCA scores

Usage

get_fpca_scores(x, components = NULL)

Arguments

Description

Get FPCA variance explained

Usage

get_fpca_variance(x)

Arguments

Description

This function downloads daily NASA POWER data for specified weather variables over a specified number of days around a given date column for multiple locations. It includes a progress bar to show the download progress.

Usage

get_nasapower(
  data,
  days_around,
  date_col,
  study_col = "study",
  pars = c("T2M", "RH2M", "PRECTOTCORR", "T2M_MAX", "T2M_MIN", "T2MDEW"),
  direction = "both"
)

Arguments

Details

The function uses the get_power function from the nasapower package to fetch weather data for a range of dates around the specified date column for each location. A progress bar is shown during the data download process, and the results are combined into a single data frame.

Value

A data frame with the downloaded weather data from NASA POWER, combined for all specified locations. Includes a variable study indicating the study identifier from the input data. Returns an empty data frame if no data is retrieved.

See Also

Other Disease modeling: get_brdwgd(), get_era5(), windowpane()

Description

The function join_count calculates spatial join count statistics for a binary matrix, identifying patterns of aggregation or randomness.

Usage

join_count(matrix_data, verbose = TRUE)

Arguments

Details

The function conducts an analysis by first counting the occurrence of specific sequences ("01 or 10" and "11" - equivalent to HD and DD) in the binary matrix. It then calculates expected values, standard deviations, and Z-scores to determine the spatial randomness or aggregation. The analysis considers both horizontal and vertical adjacency (rook case) in the matrix.

Value

A comprehensive, rich-text formatted string of results that includes:

• Statistical counts of specific binary sequences (e.g., "01 or 10", "11")

• Expected counts under the assumption of Complete Spatial Randomness (CSR)

• Standard deviations and Z-scores (ZHD for "01 or 10" sequences, ZDD for "11" sequences)

• Interpretation of whether the spatial distribution for each sequence type is "Aggregated" or "Not Aggregated" based on Z-scores

• A summary explaining the implications of these statistics and patterns

The return value aims to provide a clear understanding of the spatial arrangement's characteristics, aiding in further spatial analysis or research.

References

Madden, L. V., Hughes, G., & van den Bosch, F. (2007). The Study of Plant Disease Epidemics. The American Phytopathological Society.

See Also

Other Spatial analysis: AFSD(), BPL(), count_subareas(), count_subareas_random(), fit_gradients(), oruns_test(), oruns_test_boustrophedon(), oruns_test_byrowcol(), plot_AFSD()

Description

Perform a runs test on the input data to test for clustering or randomness.

Usage

oruns_test(x)

Arguments

Value

an r4pde.oruns object.

An r4pde.oruns object is a list containing:

• U, number of runs,

• EU, expected number of runs,

• sU, standard deviation of the expected number of runs

• Z, Z-score of the observed number of runs,

• pvalue, the p-value of the Z-score, and

• result, the test result of either "aggregation or clustering" or "randomness"

See Also

Other Spatial analysis: AFSD(), BPL(), count_subareas(), count_subareas_random(), fit_gradients(), join_count(), oruns_test_boustrophedon(), oruns_test_byrowcol(), plot_AFSD()

Examples

oruns_test(c(1, 0, 1, 1, 0, 1, 0, 0, 1, 1))

Description

Applies the ordinary runs test to a binary matrix using boustrophedon-style traversal. The function supports two modes: row-wise and column-wise boustrophedon. Each traversal flattens the matrix into a 1D sequence which is then tested using oruns_test.

Usage

oruns_test_boustrophedon(mat)

Arguments

Value

A list with two elements:

See Also

oruns_test

Other Spatial analysis: AFSD(), BPL(), count_subareas(), count_subareas_random(), fit_gradients(), join_count(), oruns_test(), oruns_test_byrowcol(), plot_AFSD()

Description

Applies the ordinary runs test to each row and column of a binary matrix individually.

Usage

oruns_test_byrowcol(mat)

Arguments

Value

A list with four elements:

See Also

oruns_test

Other Spatial analysis: AFSD(), BPL(), count_subareas(), count_subareas_random(), fit_gradients(), join_count(), oruns_test(), oruns_test_boustrophedon(), plot_AFSD()

Description

This function creates a tile plot of the foci (cluster) identified by the AFSD function. It colors each cell in a foci and labels the centroid of each cluster with the foci ID. The 'ggplot2' package is used for the plot, and will be automatically installed if not already present.

Usage

plot_AFSD(df)

Arguments

Value

A ggplot object with the scatter plot of foci (clusters).

See Also

Other Spatial analysis: AFSD(), BPL(), count_subareas(), count_subareas_random(), fit_gradients(), join_count(), oruns_test(), oruns_test_boustrophedon(), oruns_test_byrowcol()

Examples

df <- data.frame(x = sample(1:100, 500, replace = TRUE),
                 y = sample(1:100, 500, replace = TRUE),
                 i = sample(0:1, 500, replace = TRUE, prob = c(0.7, 0.3)))

# Perform the AFSD
result <- AFSD(df)
# Plot the foci
plot_AFSD(result[[3]])

Description

Plots environment-adjusted mean epidemic curves for each treatment, colored according to functional cluster membership.

The returned object is a ggplot and can be further modified using standard ggplot2 layers.

Usage

plot_curves(x, label_fun = NULL, palette = NULL, alpha = 0.9, linewidth = 1.1)

Arguments

Value

A ggplot object.

Description

Visualizes the hierarchical clustering of treatments based on functional distances among epidemic curves.

Usage

plot_dendrogram(x, label_fun = NULL, palette = NULL, show_cut = TRUE)

Arguments

Value

A ggplot object.

Description

Combines multiple diagnostic plots from diagnose_curves into a multi-panel layout.

Usage

plot_diagnostics(
  diag,
  layout = c("2x2", "vertical", "horizontal"),
  show_curve = FALSE,
  rel_heights = c(1, 0.7)
)

Arguments

Value

A composite plot object generated using cowplot.

Description

Plot method for functional instability results

Usage

plot_functional_instability(object, type = c("overall", "space", "time"), ...)

Arguments

Value

A ggplot object.

Description

Plot functional_curves

Usage

## S3 method for class 'functional_curves'
plot(x, ...)

Description

Produces a paired visualization of environment-adjusted mean curves and the corresponding functional dendrogram.

Usage

## S3 method for class 'r4pde_compare_curves'
plot(x, label_fun = NULL, palette = NULL, ...)

Arguments

Value

A patchwork object combining curves and dendrogram.

Description

Plot functional PCA results

Usage

## S3 method for class 'r4pde_functional_pca'
plot(
  x,
  type = c("scree", "components", "scores", "reconstruction", "mean"),
  components = c(1, 2),
  curve_id = NULL,
  ...
)

Arguments

Description

Print functional_curves

Usage

## S3 method for class 'functional_curves'
print(x, ...)

Description

Print functional_distances

Usage

## S3 method for class 'functional_distances'
print(x, ...)

Description

Print functional_resistance

Usage

## S3 method for class 'functional_resistance'
print(x, ...)

Description

Print method for curve diagnostics

Usage

## S3 method for class 'r4pde_curve_diagnostics'
print(x, ...)

Arguments

Value

Invisibly returns x.

Description

Print method for suggest_k

Usage

## S3 method for class 'suggest_k'
print(x, ...)

Arguments

Value

Invisibly returns x.

Description

Reconstruct curves using specified FPCA components

Usage

reconstruct_curves(x, components = NULL)

Arguments

Description

Soybean rust severity and field metadata.

Usage

RustSoybean

Format

A data frame with:

detection

Detection score or date

epidemia

Epidemic phase

latitude

Latitude

local

Location name

longitude

Longitude

planting

Planting date or stage

severity

Disease severity

Source

Del Ponte Lab internal data

Description

Simulated aggregated spatial binary disease pattern.

Usage

SpatialAggregated

Format

A data frame with:

x

x-coordinate

y

y-coordinate

Source

Simulated example

Description

Simulated random spatial binary disease pattern.

Usage

SpatialRandom

Format

A data frame with:

x

x-coordinate

y

y-coordinate

Source

Simulated example

Description

A helper function to guide selection of the k_smooth, k_trt, k_env, and gamma parameters used by functional_curves. Can infer the effective replication from a data frame or accept scalar values directly when the data are not yet available.

For sparse epidemic data it is strongly recommended to use rule = "minimum" so that k values are based on the least-replicated treatment-by-environment combination, guarding against over-fitting.

Usage

suggest_k(
  data = NULL,
  time = NULL,
  treatment = NULL,
  environment = NULL,
  n_time = NULL,
  n_env = NULL,
  rule = c("minimum", "median"),
  smoothness = c("conservative", "moderate", "flexible")
)

Arguments

Details

The function computes, for every treatment-by-environment combination, the number of unique time points at which observations are available. It then summarises these counts using the chosen rule to obtain effective_n_time. Similarly it computes, per treatment, the number of unique environments, and summarises using the same rule to obtain effective_n_env.

Recommended k values follow these heuristics:

• k_smooth: global smooth over time; capped below effective_n_time - 1.

• k_trt: treatment-specific smooth; capped below k_smooth.

• k_env: environment random effect; capped below effective_n_env.

• gamma: penalty multiplier; higher values encourage smoother fits.

Value

A named list with the following elements:

time_summary

Named numeric vector with minimum, median, and maximum unique time points per treatment-by-environment combination.

environment_summary

Named numeric vector with minimum, median, and maximum unique environments per treatment (or NULL when no environment variable is given).

effective_n_time

The effective number of time points chosen by rule.

effective_n_env

The effective number of environments chosen by rule, or NULL.

k_smooth

Recommended basis dimension for the global time smooth.

k_trt

Recommended basis dimension for the treatment-specific smooth.

k_env

Recommended basis dimension for the environment random effect.

gamma

Recommended penalisation multiplier.

message

A short interpretation message.

Examples

# Using explicit values
suggest_k(n_time = 5, n_env = 8)

# Inferring from a data frame (unquoted column names)
df <- data.frame(
  time      = rep(1:6, times = 6),
  cultivar  = rep(c("A", "B", "C"), each = 12),
  env       = rep(rep(c("E1", "E2"), each = 6), times = 3),
  severity  = runif(36, 0, 0.5)
)
suggest_k(
  data        = df,
  time        = time,
  treatment   = cultivar,
  environment = env,
  rule        = "minimum",
  smoothness  = "conservative"
)

Description

This function creates a new ggplot2 theme by modifying the cowplot::theme_half_open theme. It sets a custom font size and changes the panel background color to gray96.

Usage

theme_r4pde(font_size = 16)

Arguments

Value

A ggplot2 theme object.

Description

National dataset of white mold severity and yield.

Usage

WhiteMoldSoybean

Format

A data frame with:

country

Country name

elevation

Field elevation

elevation_class

Elevation class

harvest_year

Year of harvest

inc

Incidence

inc_check

Check plot incidence

inc_class

Incidence class

location

Location name

region

Geographical region

scl

Soybean canopy layer

season

Crop season

state

State name

study

Study identifier

treat

Treatment applied

yld

Yield

yld_check

Yield of untreated check

yld_class

Yield class

Source

Del Ponte Lab internal data

Description

This function calculates summary statistics within specified windows around a given end date in a dataset, facilitating epidemiological analysis. It allows backward, forward, or both directions of window calculations based on a user-defined variable and window lengths.

Usage

windowpane(
  data,
  end_date_col,
  date_col,
  variable,
  summary_type,
  threshold = NULL,
  window_lengths,
  direction = "backward",
  group_by_cols = NULL,
  date_format = "%Y-%m-%d"
)

Arguments

Value

A data frame with the calculated summary values for each window.

See Also

Other Disease modeling: get_brdwgd(), get_era5(), get_nasapower()

Description

This function performs bootstrapped correlation analysis for multiple predictors against a response variable. It applies the Simes method for global significance testing and calculates individual correlations, p-values, and bootstrap statistics.

Usage

windowpane_tests(
  data,
  response_var,
  corr_type = "spearman",
  R = 1000,
  global_alpha = 0.05,
  individual_alpha = 0.005
)

Arguments

Details

The function calculates correlations between the response variable and each predictor in the data frame, using bootstrapping to generate mean, standard deviation, and median estimates of the correlation. The Simes correction is applied to control for multiple testing, providing a global p-value (Pg). The function also returns the maximum observed correlation.

Value

A list containing the following elements: