Package 'serocalculator'

Title: Estimating Infection Rates from Serological Data
Description: Translates antibody levels measured in cross-sectional population samples into estimates of the frequency with which seroconversions (infections) occur in the sampled populations. Replaces the previous `seroincidence` package.
Authors: Peter Teunis [aut, cph] (Author of the method and original code.), Kristina Lai [aut, cre], Chris Orwa [aut], Kristen Aiemjoy [aut], Douglas Ezra Morrison [aut]
Maintainer: Kristina Lai <[email protected]>
License: GPL-3
Version: 1.3.0.9029
Built: 2025-02-18 04:27:47 UTC
Source: https://github.com/ucd-serg/serocalculator

Help Index

Load antibody decay curve parameter

Description

Load antibody decay curve parameter

Usage

as_curve_params(data, antigen_isos = NULL)

Arguments

data

a data.frame() or tibble::tbl_df
antigen_isos

a character() vector of antigen isotypes to be used in analyses

Value

a curve_data object (a tibble::tbl_df with extra attribute antigen_isos)

Examples

library(magrittr)
curve_data <-
  serocalculator_example("example_curve_params.csv") %>%
  read.csv() %>%
  as_curve_params()

print(curve_data)

Load noise parameters

Description

Load noise parameters

Usage

as_noise_params(data, antigen_isos = NULL)

Arguments

data

a data.frame() or tibble::tbl_df
antigen_isos

character() vector of antigen isotypes to be used in analyses

Value

a noise_params object (a tibble::tbl_df with extra attribute antigen_isos)

Examples

library(magrittr)
noise_data <-
  serocalculator_example("example_noise_params.csv") %>%
  read.csv() %>%
  as_noise_params()

print(noise_data)

Load a cross-sectional antibody survey data set

Description

Load a cross-sectional antibody survey data set

Usage

as_pop_data(
  data,
  antigen_isos = NULL,
  age = "Age",
  value = "result",
  id = "index_id",
  standardize = TRUE
)

Arguments

data

a data.frame() or tibble::tbl_df
antigen_isos

character() vector of antigen isotypes to be used in analyses
age

a character() identifying the age column
value

a character() identifying the value column
id

a character() identifying the id column
standardize

a logical() to determine standardization of columns

Value

a pop_data object (a tibble::tbl_df with extra attribute antigen_isos)

Examples

library(magrittr)
xs_data <-
  serocalculator_example("example_pop_data.csv") |>
  read.csv() |>
  as_pop_data()

print(xs_data)

graph antibody decay curves by antigen isotype

Description

graph antibody decay curves by antigen isotype

Usage

## S3 method for class 'curve_params'
autoplot(
  object,
  antigen_isos = unique(object$antigen_iso),
  ncol = min(3, length(antigen_isos)),
  ...
)

Arguments

object

a data.frame() of curve parameters (one or more MCMC samples)
antigen_isos

antigen isotypes to analyze (can subset curve_params)
ncol

how many columns of subfigures to use in panel plot
...

Arguments passed on to plot_curve_params_one_ab

verbose

verbose output

xlim

range of x values to graph

n_curves

how many curves to plot (see details).

n_points

Number of points to interpolate along the x axis (passed to ggplot2::geom_function())

rows_to_graph

which rows of curve_params to plot (overrides n_curves).

alpha

(passed to ggplot2::geom_function()) how transparent the curves should be:

  • 0 = fully transparent (invisible)

  • 1 = fully opaque

log_x

should the x-axis be on a logarithmic scale (TRUE) or linear scale (FALSE, default)?

log_y

should the Y-axis be on a logarithmic scale (default, TRUE) or linear scale (FALSE)?

Details

rows_to_graph

If you directly specify rows_to_graph when calling this function, the row numbers are enumerated separately for each antigen isotype; in other words, for the purposes of this argument, row numbers start over at 1 for each antigen isotype. There is currently no way to specify different row numbers for different antigen isotypes; if you want to do that, you will could call plot_curve_params_one_ab() directly for each antigen isotype and combine the resulting panels yourself. Or you could subset curve_params manually, before passing it to this function, and set the n_curves argument to Inf.

Value

a ggplot2::ggplot() object

Examples

library(dplyr)
library(ggplot2)
library(magrittr)

curve <-
  serocalculator_example("example_curve_params.csv") %>%
  read.csv() %>%
  as_curve_params() %>%
  filter(antigen_iso %in% c("HlyE_IgA", "HlyE_IgG")) %>%
  autoplot()

curve

Plot distribution of antibodies

Description

autoplot() method for pop_data objects

Usage

## S3 method for class 'pop_data'
autoplot(object, log = FALSE, type = "density", strata = NULL, ...)

Arguments

object

A pop_data object (from load_pop_data())
log

whether to show antibody responses on logarithmic scale
type

an option to choose type of chart: the current options are "density" or "age-scatter"
strata

the name of a variable in pop_data to stratify by (or NULL for no stratification)
...

unused

Value

a ggplot2::ggplot object

Examples

library(dplyr)
library(ggplot2)
library(magrittr)

xs_data <-
  serocalculator_example("example_pop_data.csv") %>%
  read.csv() %>%
  as_pop_data()

xs_data %>% autoplot(strata = "catchment", type = "density")
xs_data %>% autoplot(strata = "catchment", type = "age-scatter")

Plot the log-likelihood curve for the incidence rate estimate

Description

Plot the log-likelihood curve for the incidence rate estimate

Usage

## S3 method for class 'seroincidence'
autoplot(object, log_x = FALSE, ...)

Arguments

object

a seroincidence object (from est.incidence())
log_x

should the x-axis be on a logarithmic scale (TRUE) or linear scale (FALSE, default)?
...

unused

Value

a ggplot2::ggplot()

Examples

library(dplyr)
library(ggplot2)

xs_data <-
  sees_pop_data_pk_100

curve <-
  typhoid_curves_nostrat_100 %>%
  filter(antigen_iso %in% c("HlyE_IgA", "HlyE_IgG"))

noise <-
  example_noise_params_pk

est1 <- est.incidence(
  pop_data = xs_data,
  curve_param = curve,
  noise_param = noise,
  antigen_isos = c("HlyE_IgG", "HlyE_IgA"),
  build_graph = TRUE
)

# Plot the log-likelihood curve
autoplot(est1)

Plot seroincidence.by log-likelihoods

Description

Plots log-likelihood curves by stratum, for seroincidence.by objects

Usage

## S3 method for class 'seroincidence.by'
autoplot(object, ncol = min(3, length(object)), ...)

Arguments

object

a '"seroincidence.by"' object (from est.incidence.by())
ncol

number of columns to use for panel of plots
...

Arguments passed on to autoplot.seroincidence

log_x

should the x-axis be on a logarithmic scale (TRUE) or linear scale (FALSE, default)?

Value

an object of class "ggarrange", which is a ggplot2::ggplot() or a list() of ggplot2::ggplot()s.

Examples

library(dplyr)
library(ggplot2)

xs_data <-
  sees_pop_data_pk_100

curve <-
  typhoid_curves_nostrat_100 %>%
  filter(antigen_iso %in% c("HlyE_IgA", "HlyE_IgG"))

noise <-
  example_noise_params_pk

est2 <- est.incidence.by(
  strata = c("catchment"),
  pop_data = xs_data,
  curve_params = curve,
  curve_strata_varnames= NULL,
  noise_strata_varnames = NULL,
  noise_params = noise,
  antigen_isos = c("HlyE_IgG", "HlyE_IgA"),
  #num_cores = 8, #Allow for parallel processing to decrease run time
  build_graph = TRUE
)

# Plot the log-likelihood curve
autoplot(est2)

Plot method for summary.seroincidence.by objects

Description

Plot method for summary.seroincidence.by objects

Usage

## S3 method for class 'summary.seroincidence.by'
autoplot(
  object,
  xvar,
  alpha = 0.7,
  shape = 1,
  dodge_width = 0.001,
  CIs = FALSE,
  ...
)

Arguments

object

a summary.seroincidence.by object (generated by applying the summary() method to the output of est.incidence.by()).
xvar

the name of a stratifying variable in object
alpha

transparency for the points in the graph (1 = no transparency, 0 = fully transparent)
shape

shape argument for geom_point()
dodge_width

width for jitter
CIs

logical, if TRUE, add CI error bars
...

unused

Value

a ggplot2::ggplot() object

Examples

library(dplyr)
library(ggplot2)

xs_data <-
  sees_pop_data_pk_100

curve <-
  typhoid_curves_nostrat_100 %>%
  filter(antigen_iso %in% c("HlyE_IgA", "HlyE_IgG"))

noise <-
  example_noise_params_pk

est2 <- est.incidence.by(
  strata = c("catchment"),
  pop_data = xs_data,
  curve_params = curve,
  noise_params = noise,
  curve_strata_varnames= NULL,
  noise_strata_varnames = NULL,
  antigen_isos = c("HlyE_IgG", "HlyE_IgA"),
  num_cores = 2 # Allow for parallel processing to decrease run time
)

est2sum <- summary(est2)

autoplot(est2sum, "catchment")

Check the formatting of a cross-sectional antibody survey dataset.

Description

Check the formatting of a cross-sectional antibody survey dataset.

Usage

check_pop_data(pop_data, verbose = FALSE)

Arguments

pop_data

dataset to check
verbose

whether to print an "OK" message when all checks pass

Value

NULL (invisibly)

Examples

library(magrittr)

xs_data <-
  serocalculator_example("example_pop_data.csv") %>%
  read.csv() %>%
  as_pop_data()

check_pop_data(xs_data, verbose = TRUE)

Find the maximum likelihood estimate of the incidence rate parameter

Description

This function models seroincidence using maximum likelihood estimation; that is, it finds the value of the seroincidence parameter which maximizes the likelihood (i.e., joint probability) of the data.

Usage

est.incidence(
  pop_data,
  curve_params,
  noise_params,
  antigen_isos = get_biomarker_names(pop_data),
  lambda_start = 0.1,
  stepmin = 1e-08,
  stepmax = 3,
  verbose = FALSE,
  build_graph = FALSE,
  print_graph = build_graph & verbose,
  ...
)

Arguments

pop_data

a data.frame with cross-sectional serology data per antibody and age, and additional columns
curve_params

a data.frame() containing MCMC samples of parameters from the Bayesian posterior distribution of a longitudinal decay curve model. The parameter columns must be named:

  • antigen_iso: a character() vector indicating antigen-isotype combinations

  • iter: an integer() vector indicating MCMC sampling iterations

  • y0: baseline antibody level at $t=0$ ($y(t=0)$)

  • y1: antibody peak level (ELISA units)

  • t1: duration of infection

  • alpha: antibody decay rate (1/days for the current longitudinal parameter sets)

  • r: shape factor of antibody decay

noise_params

a data.frame() (or tibble::tibble()) containing the following variables, specifying noise parameters for each antigen isotype:

  • antigen_iso: antigen isotype whose noise parameters are being specified on each row

  • nu: biological noise

  • eps: measurement noise

  • y.low: lower limit of detection for the current antigen isotype

  • y.high: upper limit of detection for the current antigen isotype

antigen_isos

Character vector with one or more antibody names. Values must match pop_data
lambda_start

starting guess for incidence rate, in years/event.
stepmin

A positive scalar providing the minimum allowable relative step length.
stepmax

a positive scalar which gives the maximum allowable scaled step length. stepmax is used to prevent steps which would cause the optimization function to overflow, to prevent the algorithm from leaving the area of interest in parameter space, or to detect divergence in the algorithm. stepmax would be chosen small enough to prevent the first two of these occurrences, but should be larger than any anticipated reasonable step.
verbose

logical: if TRUE, print verbose log information to console
build_graph

whether to graph the log-likelihood function across a range of incidence rates (lambda values)
print_graph

whether to display the log-likelihood curve graph in the course of running est.incidence()
...

Arguments passed on to stats::nlm

typsize

an estimate of the size of each parameter at the minimum.

fscale

an estimate of the size of f at the minimum.

ndigit

the number of significant digits in the function f.

gradtol

a positive scalar giving the tolerance at which the scaled gradient is considered close enough to zero to terminate the algorithm. The scaled gradient is a measure of the relative change in f in each direction p[i] divided by the relative change in p[i].

iterlim

a positive integer specifying the maximum number of iterations to be performed before the program is terminated.

check.analyticals

a logical scalar specifying whether the analytic gradients and Hessians, if they are supplied, should be checked against numerical derivatives at the initial parameter values. This can help detect incorrectly formulated gradients or Hessians.

Value

a "seroincidence" object, which is a stats::nlm() fit object with extra meta-data attributes lambda_start, antigen_isos, and ll_graph

Examples

library(dplyr)

xs_data <-
  sees_pop_data_pk_100

curve <-
  typhoid_curves_nostrat_100 %>%
  filter(antigen_iso %in% c("HlyE_IgA", "HlyE_IgG"))

noise <-
  example_noise_params_pk

est1 <- est.incidence(
  pop_data = xs_data,
  curve_params = curve,
  noise_params = noise,
  antigen_isos = c("HlyE_IgG", "HlyE_IgA"),
)

summary(est1)

Estimate Seroincidence

Description

Function to estimate seroincidences based on cross-sectional serology data and longitudinal response model.

Usage

est.incidence.by(
  pop_data,
  curve_params,
  noise_params,
  strata,
  curve_strata_varnames = strata,
  noise_strata_varnames = strata,
  antigen_isos = pop_data %>% pull("antigen_iso") %>% unique(),
  lambda_start = 0.1,
  build_graph = FALSE,
  num_cores = 1L,
  verbose = FALSE,
  print_graph = FALSE,
  ...
)

Arguments

pop_data

a data.frame with cross-sectional serology data per antibody and age, and additional columns corresponding to each element of the strata input
curve_params

a data.frame() containing MCMC samples of parameters from the Bayesian posterior distribution of a longitudinal decay curve model. The parameter columns must be named:

  • antigen_iso: a character() vector indicating antigen-isotype combinations

  • iter: an integer() vector indicating MCMC sampling iterations

  • y0: baseline antibody level at $t=0$ ($y(t=0)$)

  • y1: antibody peak level (ELISA units)

  • t1: duration of infection

  • alpha: antibody decay rate (1/days for the current longitudinal parameter sets)

  • r: shape factor of antibody decay

noise_params

a data.frame() (or tibble::tibble()) containing the following variables, specifying noise parameters for each antigen isotype:

  • antigen_iso: antigen isotype whose noise parameters are being specified on each row

  • nu: biological noise

  • eps: measurement noise

  • y.low: lower limit of detection for the current antigen isotype

  • y.high: upper limit of detection for the current antigen isotype

strata

a character vector of stratum-defining variables. Values must be variable names in pop_data.
curve_strata_varnames

A subset of strata. Values must be variable names in curve_params. Default = "".
noise_strata_varnames

A subset of strata. Values must be variable names in noise_params. Default = "".
antigen_isos

Character vector with one or more antibody names. Values must match pop_data
lambda_start

starting guess for incidence rate, in years/event.
build_graph

whether to graph the log-likelihood function across a range of incidence rates (lambda values)
num_cores

Number of processor cores to use for calculations when computing by strata. If set to more than 1 and package parallel is available, then the computations are executed in parallel. Default = 1L.
verbose

logical: if TRUE, print verbose log information to console
print_graph

whether to display the log-likelihood curve graph in the course of running est.incidence()
...

Arguments passed on to est.incidence, stats::nlm

stepmin

A positive scalar providing the minimum allowable relative step length.

stepmax

a positive scalar which gives the maximum allowable scaled step length. stepmax is used to prevent steps which would cause the optimization function to overflow, to prevent the algorithm from leaving the area of interest in parameter space, or to detect divergence in the algorithm. stepmax would be chosen small enough to prevent the first two of these occurrences, but should be larger than any anticipated reasonable step.

typsize

an estimate of the size of each parameter at the minimum.

fscale

an estimate of the size of f at the minimum.

ndigit

the number of significant digits in the function f.

gradtol

a positive scalar giving the tolerance at which the scaled gradient is considered close enough to zero to terminate the algorithm. The scaled gradient is a measure of the relative change in f in each direction p[i] divided by the relative change in p[i].

iterlim

a positive integer specifying the maximum number of iterations to be performed before the program is terminated.

check.analyticals

a logical scalar specifying whether the analytic gradients and Hessians, if they are supplied, should be checked against numerical derivatives at the initial parameter values. This can help detect incorrectly formulated gradients or Hessians.

Details

If strata is left empty, a warning will be produced, recommending that you use est.incidence() for unstratified analyses, and then the data will be passed to est.incidence(). If for some reason you want to use est.incidence.by() with no strata instead of calling est.incidence(), you may use NA, NULL, or "" as the strata argument to avoid that warning.

Value

  • if strata has meaningful inputs: An object of class "seroincidence.by"; i.e., a list of "seroincidence" objects from est.incidence(), one for each stratum, with some meta-data attributes.

  • if strata is missing, NULL, NA, or "": An object of class "seroincidence".

Examples

library(dplyr)

xs_data <-
  sees_pop_data_pk_100

curve <-
  typhoid_curves_nostrat_100 %>%
  filter(antigen_iso %in% c("HlyE_IgA", "HlyE_IgG"))

noise <-
  example_noise_params_pk

est2 <- est.incidence.by(
  strata = "catchment",
  pop_data = xs_data,
  curve_params = curve,
  noise_params = noise,
  antigen_isos = c("HlyE_IgG", "HlyE_IgA"),
  # num_cores = 8 # Allow for parallel processing to decrease run time
  iterlim = 5 # limit iterations for the purpose of this example
)
print(est2)
summary(est2)

Small example of noise parameters for typhoid

Description

A subset of noise parameter estimates from the SEES study, for examples and testing, for Pakistan

Usage

example_noise_params_pk

Format

example_noise_params_pk

A curve_params object (from as_curve_params()) with 4 rows and 7 columns:

antigen_iso

which antigen and isotype are being measured (data is in long format)

Country

Location for which the noise parameters were estimated

y.low

Lower limit of detection

eps

Measurement noise, defined by a CV (coefficient of variation) as the ratio of the standard deviation to the mean for replicates. Note that the CV should ideally be measured across plates rather than within the same plate.

nu

Biological noise: error from cross-reactivity to other antibodies. It is defined as the 95th percentile of the distribution of antibody responses to the antigen-isotype in a population with no exposure.

y.high

Upper limit of detection

Lab

Lab for which noise was estimated.

Source

https://osf.io/rtw5k

Small example of noise parameters for typhoid

Description

A subset of noise parameter estimates from the SEES study, for examples and testing.

Usage

example_noise_params_sees

Format

example_noise_params_pk

A curve_params object (from as_curve_params()) with 4 rows and 7 columns:

antigen_iso

which antigen and isotype are being measured (data is in long format)

Country

Location for which the noise parameters were estimated

y.low

Lower limit of detection

eps

Measurement noise, defined by a CV (coefficient of variation) as the ratio of the standard deviation to the mean for replicates. Note that the CV should ideally be measured across plates rather than within the same plate.

nu

Biological noise: error from cross-reactivity to other antibodies. It is defined as the 95th percentile of the distribution of antibody responses to the antigen-isotype in a population with no exposure.

y.high

Upper limit of detection

Lab

Lab for which noise was estimated.

Source

https://osf.io/rtw5k

Extract biomarker levels

Description

Extract biomarker levels

Usage

get_biomarker_levels(object, ...)

Arguments

object

a pop_data object
...

unused

Value

the biomarker levels in object

Examples

sees_pop_data_100 |> get_biomarker_levels()

Get biomarker variable name

Description

Get biomarker variable name

Usage

get_biomarker_names_var(object, ...)

Arguments

object

a pop_data object
...

unused

Value

a character string identifying the biomarker names column in object

Examples

sees_pop_data_100 |> get_biomarker_names_var()

Get antibody measurement values

Description

Get antibody measurement values

Usage

get_values(object, ...)

Arguments

object

a pop_data object
...

unused

Value

a numeric vector of antibody measurement values

Examples

sees_pop_data_100 |> get_values()

Extract antibody measurement values

Description

Extract antibody measurement values

Usage

get_values_var(object, ...)

Arguments

object

a pop_data object
...

unused

Value

the name of the column in object specified as containing antibody abundance measurements

Examples

sees_pop_data_100 |> get_values_var()

Graph log-likelihood of data

Description

Graph log-likelihood of data

Usage

graph_loglik(
  pop_data,
  curve_params,
  noise_params,
  antigen_isos = pop_data %>% get_biomarker_levels(),
  x = 10^seq(-3, 0, by = 0.1),
  highlight_points = NULL,
  highlight_point_names = "highlight_points",
  log_x = FALSE,
  previous_plot = NULL,
  curve_label = paste(antigen_isos, collapse = " + "),
  ...
)

Arguments

pop_data

a data.frame() with cross-sectional serology data by antibody and age, and additional columns
curve_params

a data.frame() containing MCMC samples of parameters from the Bayesian posterior distribution of a longitudinal decay curve model. The parameter columns must be named:

  • antigen_iso: a character() vector indicating antigen-isotype combinations

  • iter: an integer() vector indicating MCMC sampling iterations

  • y0: baseline antibody level at $t=0$ ($y(t=0)$)

  • y1: antibody peak level (ELISA units)

  • t1: duration of infection

  • alpha: antibody decay rate (1/days for the current longitudinal parameter sets)

  • r: shape factor of antibody decay

noise_params

a data.frame() (or tibble::tibble()) containing the following variables, specifying noise parameters for each antigen isotype:

  • antigen_iso: antigen isotype whose noise parameters are being specified on each row

  • nu: biological noise

  • eps: measurement noise

  • y.low: lower limit of detection for the current antigen isotype

  • y.high: upper limit of detection for the current antigen isotype

antigen_isos

Character vector listing one or more antigen isotypes. Values must match pop_data.
x

sequence of lambda values to graph
highlight_points

a possible highlighted value
highlight_point_names

labels for highlighted points
log_x

should the x-axis be on a logarithmic scale (TRUE) or linear scale (FALSE, default)?
previous_plot

if not NULL, the current data is added to the existing graph
curve_label

if not NULL, add a label for the curve
...

Arguments passed on to log_likelihood

verbose

logical: if TRUE, print verbose log information to console

Value

a ggplot2::ggplot()

Examples

library(dplyr)
library(tibble)

# Load cross-sectional data
xs_data <-
  sees_pop_data_pk_100

# Load curve parameters and subset for the purposes of this example
curve <-
  typhoid_curves_nostrat_100 %>%
  filter(antigen_iso %in% c("HlyE_IgA", "HlyE_IgG"))

# Load noise parameters
cond <- tibble(
  antigen_iso = c("HlyE_IgG", "HlyE_IgA"),
  nu = c(0.5, 0.5),                          # Biologic noise (nu)
  eps = c(0, 0),                             # M noise (eps)
  y.low = c(1, 1),                           # Low cutoff (llod)
  y.high = c(5e6, 5e6))                      # High cutoff (y.high)

# Graph the log likelihood
lik_HlyE_IgA <- # nolint: object_name_linter
  graph_loglik(
    pop_data = xs_data,
    curve_params = curve,
    noise_params = cond,
    antigen_isos = "HlyE_IgA",
    log_x = TRUE
)

lik_HlyE_IgA # nolint: object_name_linter

Graph estimated antibody decay curve

Description

Graph estimated antibody decay curve

Usage

graph.curve.params(
  curve_params,
  antigen_isos = unique(curve_params$antigen_iso),
  verbose = FALSE,
  show_quantiles = TRUE,
  show_all_curves = FALSE,
  alpha_samples = 0.3
)

Arguments

curve_params

a data.frame() containing MCMC samples of antibody decay curve parameters
antigen_isos

antigen isotypes
verbose

verbose output
show_quantiles

whether to show point-wise (over time) quantiles
show_all_curves

whether to show individual curves under quantiles
alpha_samples

alpha parameter passed to ggplot2::geom_line (has no effect if show_all_curves = FALSE)

Value

a ggplot2::ggplot() object

Examples

curve <-
  typhoid_curves_nostrat_100 |>
  dplyr::filter(antigen_iso %in% c("HlyE_IgA", "HlyE_IgG"))

plot1 <- graph.curve.params(curve)

print(plot1)

plot2 <- graph.curve.params(curve, show_all_curves = TRUE)
show(plot2)

Load antibody decay curve parameter samples

Description

Load antibody decay curve parameter samples

Usage

load_curve_params(file_path, antigen_isos = NULL)

Arguments

file_path

path to an RDS file containing MCMC samples of antibody decay curve parameters y0, y1, t1, alpha, and r, stored as a data.frame() or tibble::tbl_df
antigen_isos

character() vector of antigen isotypes to be used in analyses

Value

a curve_params object (a tibble::tbl_df with extra attribute antigen_isos)

Examples

curve <- load_curve_params(serocalculator_example("example_curve_params.rds"))

print(curve)

Load noise parameters

Description

Load noise parameters

Usage

load_noise_params(file_path, antigen_isos = NULL)

Arguments

file_path

path to an RDS file containing biologic and measurement noise of antibody decay curve parameters y.low, eps, nu, and y.high, stored as a data.frame() or tibble::tbl_df
antigen_isos

character() vector of antigen isotypes to be used in analyses

Value

a noise object (a tibble::tbl_df with extra attribute antigen_isos)

Examples

noise <- load_noise_params(serocalculator_example("example_noise_params.rds"))
print(noise)

Load a cross-sectional antibody survey data set

Description

Load a cross-sectional antibody survey data set

Usage

load_pop_data(file_path, ...)

Arguments

file_path

path to an RDS file containing a cross-sectional antibody survey data set, stored as a data.frame() or tibble::tbl_df
...

Arguments passed on to as_pop_data

data

a data.frame() or tibble::tbl_df

antigen_isos

character() vector of antigen isotypes to be used in analyses

age

a character() identifying the age column

id

a character() identifying the id column

value

a character() identifying the value column

standardize

a logical() to determine standardization of columns

Value

a pop_data object (a tibble::tbl_df with extra attributes)

Examples

xs_data <- load_pop_data(serocalculator_example("example_pop_data.rds"))

print(xs_data)

Calculate log-likelihood

Description

Calculates the log-likelihood of a set of cross-sectional antibody response data, for a given incidence rate (lambda) value.

Usage

log_likelihood(
  lambda,
  pop_data,
  curve_params,
  noise_params,
  antigen_isos = get_biomarker_levels(pop_data),
  verbose = FALSE,
  ...
)

Arguments

lambda

a numeric vector of incidence parameters, in events per person-year
pop_data

a data.frame() with cross-sectional serology data by antibody and age, and additional columns
curve_params

a data.frame() containing MCMC samples of parameters from the Bayesian posterior distribution of a longitudinal decay curve model. The parameter columns must be named:

  • antigen_iso: a character() vector indicating antigen-isotype combinations

  • iter: an integer() vector indicating MCMC sampling iterations

  • y0: baseline antibody level at $t=0$ ($y(t=0)$)

  • y1: antibody peak level (ELISA units)

  • t1: duration of infection

  • alpha: antibody decay rate (1/days for the current longitudinal parameter sets)

  • r: shape factor of antibody decay

noise_params

a data.frame() (or tibble::tibble()) containing the following variables, specifying noise parameters for each antigen isotype:

  • antigen_iso: antigen isotype whose noise parameters are being specified on each row

  • nu: biological noise

  • eps: measurement noise

  • y.low: lower limit of detection for the current antigen isotype

  • y.high: upper limit of detection for the current antigen isotype

antigen_isos

Character vector listing one or more antigen isotypes. Values must match pop_data.
verbose

logical: if TRUE, print verbose log information to console
...

additional arguments passed to other functions (not currently used).

Value

the log-likelihood of the data with the current parameter values

Examples

library(dplyr)
library(tibble)

# Load cross-sectional data
xs_data <-
  sees_pop_data_pk_100

# Load curve parameters and subset for the purposes of this example
curve <-
  typhoid_curves_nostrat_100 %>%
  filter(antigen_iso %in% c("HlyE_IgA", "HlyE_IgG"))

# Load noise params
cond <- tibble(
  antigen_iso = c("HlyE_IgG", "HlyE_IgA"),
  nu = c(0.5, 0.5), # Biologic noise (nu)
  eps = c(0, 0), # M noise (eps)
  y.low = c(1, 1), # low cutoff (llod)
  y.high = c(5e6, 5e6)
) # high cutoff (y.high)

# Calculate log-likelihood
ll_AG <- log_likelihood(
  pop_data = xs_data,
  curve_params = curve,
  noise_params = cond,
  antigen_isos = c("HlyE_IgG", "HlyE_IgA"),
  lambda = 0.1
) %>% print()

Small example cross-sectional data set

Description

A subset of data from the SEES data, for examples and testing.

Usage

sees_pop_data_100

Format

sees_pop_data_pk_100

A pop_data object (from as_pop_data()) with 200 rows and 8 columns:

id

Observation ID

Country

Country where the participant was living

cluster

survey sampling cluster

catchment

survey catchment area

age

participant's age when sampled, in years

antigen_iso

which antigen and isotype are being measured (data is in long format)

value

concentration of antigen isotype, in ELISA units

Source

https://osf.io/n6cp3

Small example cross-sectional data set

Description

A subset of data from the SEES data, for examples and testing, data from Pakistan only.

Usage

sees_pop_data_pk_100

Format

sees_pop_data_pk_100

A pop_data object (from as_pop_data()) with 200 rows and 8 columns:

id

Observation ID

Country

Country where the participant was living

cluster

survey sampling cluster

catchment

survey catchment area

age

participant's age when sampled, in years

antigen_iso

which antigen and isotype are being measured (data is in long format)

value

concentration of antigen isotype, in ELISA units

Source

https://osf.io/n6cp3

Get path to an example file

Description

The serocalculator package comes bundled with a number of sample files in its inst/extdata directory. This serocalculator_example() function make those sample files easy to access.

Usage

serocalculator_example(file = NULL)

Arguments

file

Name of file. If NULL, the example files will be listed.

Details

Adapted from readr::readr_example() following the guidance in https://r-pkgs.org/data.html#sec-data-example-path-helper.

Value

a character string providing the path to the file specified by file, or a vector or available files if file = NULL.

Examples

serocalculator_example()
serocalculator_example("example_pop_data.csv")

Simulate a cross-sectional serosurvey with noise

Description

Makes a cross-sectional data set (age, y(t) set) and adds noise, if desired.

Usage

sim_pop_data(
  lambda = 0.1,
  n_samples = 100,
  age_range = c(0, 20),
  age_fixed = NA,
  antigen_isos = intersect(get_biomarker_levels(curve_params), rownames(noise_limits)),
  n_mcmc_samples = 0,
  renew_params = FALSE,
  add_noise = FALSE,
  curve_params,
  noise_limits,
  format = "wide",
  verbose = FALSE,
  ...
)

Arguments

lambda

a numeric() scalar indicating the incidence rate (in events per person-years)
n_samples

number of samples to simulate
age_range

age range of sampled individuals, in years
age_fixed

specify the curve parameters to use by age (does nothing at present?)
antigen_isos

Character vector with one or more antibody names. Values must match curve_params.
n_mcmc_samples

how many MCMC samples to use:

  • when n_mcmc_samples is in 1:4000 a fixed posterior sample is used

  • when n_mcmc_samples = 0, a random sample is chosen

renew_params

whether to generate a new parameter set for each infection

  • renew_params = TRUE generates a new parameter set for each infection

  • renew_params = FALSE keeps the one selected at birth, but updates baseline y0

add_noise

a logical() indicating whether to add biological and measurement noise
curve_params

a data.frame() containing MCMC samples of parameters from the Bayesian posterior distribution of a longitudinal decay curve model. The parameter columns must be named:

  • antigen_iso: a character() vector indicating antigen-isotype combinations

  • iter: an integer() vector indicating MCMC sampling iterations

  • y0: baseline antibody level at $t=0$ ($y(t=0)$)

  • y1: antibody peak level (ELISA units)

  • t1: duration of infection

  • alpha: antibody decay rate (1/days for the current longitudinal parameter sets)

  • r: shape factor of antibody decay

noise_limits

biologic noise distribution parameters
format

a character() variable, containing either:

  • "long" (one measurement per row) or

  • "wide" (one serum sample per row)

verbose

logical: if TRUE, print verbose log information to console
...

Arguments passed on to simcs.tinf

Value

a tibble::tbl_df containing simulated cross-sectional serosurvey data, with columns:

  • age: age (in days)

  • one column for each element in the antigen_iso input argument

Examples

# Load curve parameters
dmcmc <- typhoid_curves_nostrat_100

# Specify the antibody-isotype responses to include in analyses
antibodies <- c("HlyE_IgA", "HlyE_IgG")

# Set seed to reproduce results
set.seed(54321)

# Simulated incidence rate per person-year
lambda <- 0.2
# Range covered in simulations
lifespan <- c(0, 10)
# Cross-sectional sample size
nrep <- 100

# Biologic noise distribution
dlims <- rbind(
  "HlyE_IgA" = c(min = 0, max = 0.5),
  "HlyE_IgG" = c(min = 0, max = 0.5)
)

# Generate cross-sectional data
csdata <- sim_pop_data(
  curve_params = dmcmc,
  lambda = lambda,
  n_samples = nrep,
  age_range = lifespan,
  antigen_isos = antibodies,
  n_mcmc_samples = 0,
  renew_params = TRUE,
  add_noise = TRUE,
  noise_limits = dlims,
  format = "long"
)

Simulate multiple data sets

Description

Simulate multiple data sets

Usage

sim_pop_data_multi(
  nclus = 10,
  lambdas = c(0.05, 0.1, 0.15, 0.2, 0.3),
  num_cores = max(1, parallel::detectCores() - 1),
  rng_seed = 1234,
  verbose = FALSE,
  ...
)

Arguments

nclus

number of clusters
lambdas

#incidence rate, in events/person*year
num_cores

number of cores to use for parallel computations
rng_seed

starting seed for random number generator, passed to rngtools::RNGseq()
verbose

whether to report verbose information
...

Arguments passed on to sim_pop_data

lambda

a numeric() scalar indicating the incidence rate (in events per person-years)

n_samples

number of samples to simulate

age_range

age range of sampled individuals, in years

age_fixed

specify the curve parameters to use by age (does nothing at present?)

antigen_isos

Character vector with one or more antibody names. Values must match curve_params.

n_mcmc_samples

how many MCMC samples to use:

  • when n_mcmc_samples is in 1:4000 a fixed posterior sample is used

  • when n_mcmc_samples = 0, a random sample is chosen

renew_params

whether to generate a new parameter set for each infection

  • renew_params = TRUE generates a new parameter set for each infection

  • renew_params = FALSE keeps the one selected at birth, but updates baseline y0

add_noise

a logical() indicating whether to add biological and measurement noise

noise_limits

biologic noise distribution parameters

format

a character() variable, containing either:

  • "long" (one measurement per row) or

  • "wide" (one serum sample per row)

curve_params

a data.frame() containing MCMC samples of parameters from the Bayesian posterior distribution of a longitudinal decay curve model. The parameter columns must be named:

  • antigen_iso: a character() vector indicating antigen-isotype combinations

  • iter: an integer() vector indicating MCMC sampling iterations

  • y0: baseline antibody level at $t=0$ ($y(t=0)$)

  • y1: antibody peak level (ELISA units)

  • t1: duration of infection

  • alpha: antibody decay rate (1/days for the current longitudinal parameter sets)

  • r: shape factor of antibody decay

Value

a tibble::tibble()

Examples

# Load curve parameters
dmcmc <- typhoid_curves_nostrat_100

# Specify the antibody-isotype responses to include in analyses
antibodies <- c("HlyE_IgA", "HlyE_IgG")

# Set seed to reproduce results
set.seed(54321)

# Simulated incidence rate per person-year
lambdas = c(.05, .1, .15, .2, .3)

# Range covered in simulations
lifespan <- c(0, 10);

# Cross-sectional sample size
nrep <- 100

# Biologic noise distribution
dlims <- rbind(
  "HlyE_IgA" = c(min = 0, max = 0.5),
  "HlyE_IgG" = c(min = 0, max = 0.5)
)

sim_pop_data_multi(
  curve_params = dmcmc,
  lambdas = lambdas,
  n_samples = nrep,
  age_range = lifespan,
  antigen_isos = antibodies,
  n_mcmc_samples = 0,
  renew_params = TRUE,
  add_noise = TRUE,
  noise_limits = dlims,
  format = "long",
  nclus = 10)

Extract strata from an object

Description

Generic method for extracting strata from objects. See strata.seroincidence.by()

Usage

strata(x)

Arguments

x

an object

Value

the strata of x

Summarizing fitted seroincidence models

Description

This function is a summary() method for seroincidence objects.

Usage

## S3 method for class 'seroincidence'
summary(object, coverage = 0.95, verbose = TRUE, ...)

Arguments

object

a list(), outputted by stats::nlm() or est.incidence()
coverage

desired confidence interval coverage probability
verbose

whether to produce verbose messaging
...

unused

Value

a tibble::tibble() containing the following:

  • est.start: the starting guess for incidence rate

  • ageCat: the age category we are analyzing

  • incidence.rate: the estimated incidence rate, per person year

  • CI.lwr: lower limit of confidence interval for incidence rate

  • CI.upr: upper limit of confidence interval for incidence rate

  • coverage: coverage probability

  • log.lik: log-likelihood of the data used in the call to est.incidence(), evaluated at the maximum-likelihood estimate of lambda (i.e., at incidence.rate)

  • iterations: the number of iterations used

  • antigen_isos: a list of antigen isotypes used in the analysis

  • nlm.convergence.code: information about convergence of the likelihood maximization procedure performed by nlm() (see "Value" section of stats::nlm(), component code); codes 3-5 indicate issues:

    • 1: relative gradient is close to zero, current iterate is probably solution.

    • 2: successive iterates within tolerance, current iterate is probably solution.

    • 3: Last global step failed to locate a point lower than x. Either x is an approximate local minimum of the function, the function is too non-linear for this algorithm, or stepmin in est.incidence() (a.k.a., steptol in stats::nlm()) is too large.

    • 4: iteration limit exceeded; increase iterlim.

    • 5: maximum step size stepmax exceeded five consecutive times. Either the function is unbounded below, becomes asymptotic to a finite value from above in some direction, or stepmax is too small.

Examples

library(dplyr)

xs_data <-
  sees_pop_data_pk_100

curve <-
  typhoid_curves_nostrat_100 |>
  filter(antigen_iso %in% c("HlyE_IgA", "HlyE_IgG"))

noise <-
  example_noise_params_pk

est1 <- est.incidence(
  pop_data = xs_data,
  curve_params = curve,
  noise_params = noise,
  antigen_isos = c("HlyE_IgG", "HlyE_IgA")
)

summary(est1)

Summary Method for "seroincidence.by" Objects

Description

Calculate seroincidence from output of the seroincidence calculator est.incidence.by().

Usage

## S3 method for class 'seroincidence.by'
summary(
  object,
  confidence_level = 0.95,
  show_deviance = TRUE,
  show_convergence = TRUE,
  verbose = FALSE,
  ...
)

Arguments

object

A dataframe containing output of function est.incidence.by().
confidence_level

desired confidence interval coverage probability
show_deviance

Logical flag (FALSE/TRUE) for reporting deviance (-2*log(likelihood) at estimated seroincidence. Default = TRUE.
show_convergence

Logical flag (FALSE/TRUE) for reporting convergence (see help for optim() for details). Default = FALSE.
verbose

a logical scalar indicating whether to print verbose messages to the console
...

Additional arguments affecting the summary produced.

Value

A summary.seroincidence.by object, which is a tibble::tibble, with the following columns:

  • incidence.rate maximum likelihood estimate of lambda (seroincidence)

  • CI.lwr lower confidence bound for lambda

  • CI.upr upper confidence bound for lambda

  • Deviance (included if show_deviance = TRUE) Negative log likelihood (NLL) at estimated (maximum likelihood) lambda)

  • nlm.convergence.code (included if show_convergence = TRUE) Convergence information returned by stats::nlm()

The object also has the following metadata (accessible through base::attr()):

Examples

library(dplyr)

xs_data <-
  sees_pop_data_pk_100

curve <-
  typhoid_curves_nostrat_100 |>
  filter(antigen_iso %in% c("HlyE_IgA", "HlyE_IgG"))

noise <-
  example_noise_params_pk

# estimate seroincidence
est2 <- est.incidence.by(
  strata = c("catchment"),
  pop_data = xs_data,
  curve_params = curve,
  noise_params = noise,
  antigen_isos = c("HlyE_IgG", "HlyE_IgA"),
  # num_cores = 8 # Allow for parallel processing to decrease run time
)

# calculate summary statistics for the seroincidence object
summary(est2)

Small example of antibody response curve parameters for typhoid

Description

A subset of data from the SEES study, for examples and testing.

Usage

typhoid_curves_nostrat_100

Format

typhoid_curves_nostrat_100

A curve_params object (from as_curve_params()) with 500 rows and 7 columns:

antigen_iso

which antigen and isotype are being measured (data is in long format)

iter

MCMC iteration

y0

Antibody concentration at t = 0 (start of active infection)

y1

Antibody concentration at t = t1 (end of active infection)

t1

Duration of active infection

alpha

Antibody decay rate coefficient

r

Antibody decay rate exponent parameter

Source

https://osf.io/rtw5k