Package 'Rcurvep'

Calculate the knee point on the exponential-like curve

Description

Currently two methods have been implemented to get the "keen-point" from the variance(y) - threshold(x) curve. One is to use the original y values to draw a straight line between the lowest x value (p1) to highest x value (p2). The knee-point is the x that has the longest distance to the line. The other one is to fit the data first then use the fitted responses to do the same analysis. Currently the first method is preferred.

Usage

cal_knee_point(d, xaxis, yaxis, p1 = NULL, p2 = NULL, plot = TRUE)

Arguments

d	A tibble.
xaxis	The column name in the d to be the x-axis in the exponential-like curve
yaxis	The column name in the d to be the y-axis in the exponential-like curve
p1	Default = NULL, or an integer value to manually set the first index of line.
p2	Default = NULL, or an integer value to manually set the last index of line.
plot	Default = TRUE, plot the diagnostic plot.

Value

A list with two components: stats and outcome.

• stats: a tibble, including pooled variance (pvar), fitted responses (y_exp_fit, y_lm_fit), distance to the line (dist2l)

• outcome: a tibble, including estimated BMRs (bmr)

; Suffix in the stats and outcome tibble: "ori" (original values), "exp"(exponential fit). prefix in the outcome tibble, "cor" (correlation between the fitted responses and the original responses), "bmr" (benchmark response), "qc" (quality control).

See Also

estimate_dataset_bmr()

Examples

inp <- data.frame(
x = seq(5, 95, by = 5),
y = c(0.0537, 0.0281, 0.0119, 0.0109, 0.0062, 0.0043, 0.0043, 0.0042,
0.0041, 0.0043, 0.0044, 0.0044, 0.0046, 0.0051,
0.0055, 0.0057, 0.0072, 0.0068, 0.0035)
)

out <- cal_knee_point(inp,"x", "y", plot = FALSE)
plot(out)

---

Run Curvep on datasets of concentration-response data with a combination of Curvep parameters

Description

It simplifies the steps of run_rcurvep() by wrapping the create_dataset() in the function.

Usage

combi_run_rcurvep(
  d,
  n_samples = NULL,
  vdata = NULL,
  mask = 0,
  keep_sets = c("act_set", "resp_set", "fp_set"),
  ...
)

Arguments

d	Datasets with concentration-response data. Examples are zfishbeh and zfishdev.
n_samples	NULL (default) for not to simulate responses or an integer number to indicate the number of responses per concentration to simulate.
vdata	NULL (default) for not to simulate responses or a vector of numeric responses in vehicle control wells to use as error. This parameter only works when n_samples is not NULL; an experimental feature.
mask	Default = 0, for no mask (values in the mask column all 0). Use a vector of integers to mask the responses: 1 to mask the response at the highest concentration; 2 to mask the response at the second highest concentration, and so on. If mask column exists, the setting will be ignored.
keep_sets	The types of output to be reported. Allowed values: act_set, resp_set, fp_set. Multiple values are allowed. act_set is the must. act_set: activity data resp_set: response data fp_set: fingerprint data
...	Curvep settings. See curvep_defaults() for allowed parameters. These can be used to overwrite the default values.

Value

An rcurvep object. It has two components: result, config The result component is also a list of output sets depending on the parameter, keep_sets. The config component is a curvep_config object.

Often used columns in the act_set: AUC (area under the curve), wAUC (weighted AUC), POD (point-of-departure), EC50 (Half maximal effective concentration), nCorrected (number of corrected points).

See Also

run_rcurvep() summarize_rcurvep_output()

Examples

data(zfishbeh)

# 2 simulated sample curves +
# using two thresholds +
# mask the response at the higest concentration
# only to output the act_set

out <- combi_run_rcurvep(
  zfishbeh,
  n_samples = 2,
  TRSH = c(5, 10),
  mask = 1,
  keep_sets = "act_set")

# create the zfishdev_act dataset


 data(zfishdev_all)
 zfishdev_act <- combi_run_rcurvep(
   zfishdev_all, n_samples = 100, keep_sets = c("act_set"),TRSH = seq(5, 95, by = 5),
   RNGE = 1000000, CARR = 20, seed = 300
 )

---

Create concentration-response datasets that can be applied in the run_rcurvep()

Description

The input dataset is created either by summarizing the response data or by simulating the response data.

Usage

create_dataset(d, n_samples = NULL, vdata = NULL)

Arguments

d	Datasets with concentration-response data. Examples are zfishbeh and zfishdev.
n_samples	NULL (default) for not to simulate responses or an integer number to indicate the number of responses per concentration to simulate.
vdata	NULL (default) for not to simulate responses or a vector of numeric responses in vehicle control wells to use as error. This parameter only works when n_samples is not NULL; an experimental feature.

Details

Curvep requires 1-to-1 concentration response relationship. For the dataset that does not meet the requirement, the following strategies are applied:

Summary (when n_samples = NULL)

• For dichotomous responses, percentage is reported (n_in/N*100).

• For continuous responses, median value of responses per concentration is reported.

Simulation (when n_samples is a positive integer)

• For dichotomous responses, bootstrap approach is used on the "n_in" vector to create a vector of percent response.

• For continuous responses, options are a) direct sampling; b) responses from the linear fit using the original data + error of responses based on the supplied vehicle control data

Value

The original dataset with a new column, sample_id (if n_samples is not NULL) or the summarized dataset with columns as zfishbeh.

See Also

run_rcurvep()

Examples

# datasets with continuous response data
data(zfishbeh)

## default
d <- create_dataset(zfishbeh)

## add samples
d <- create_dataset(zfishbeh, n_samples = 3)

## add samples and vdata
d <- create_dataset(zfishbeh, n_samples = 3, vdata = rnorm(100))

# dataset with dichotomous response data
data(zfishdev)

## default
d <- create_dataset(zfishdev)

## add samples
d <- create_dataset(zfishdev, n_samples = 3)

---

The Curvep function to process one set of concentration-response data

Description

The relationship between concentration and response has to be 1 to 1. The function is the backbone of run_rcurvep() and combi_run_rcurvep().

Usage

curvep(
  Conc,
  Resp,
  Mask = NULL,
  TRSH = 15,
  RNGE = -100,
  MXDV = 5,
  CARR = 0,
  BSFT = 3,
  USHP = 4,
  TrustHi = FALSE,
  StrictImp = TRUE,
  DUMV = -999,
  TLOG = -24,
  ...
)

Arguments

Conc	Array of concentrations, e.g., in Molar units, can be log-transformed, in which case internal log-transformation is skipped.
Resp	Array of responses at corresponding concentrations, e.g., raw measurements or normalized to controls.
Mask	array of 1/0 flags indicating invalidated measurements (default = NULL).
TRSH	Base(zero-)line threshold (default = 15).
RNGE	Target range of responses (default = -100).
MXDV	Maximum allowed deviation from monotonicity (default = 5).
CARR	Carryover detection threshold (default = 0, analysis skipped if set to 0)
BSFT	For baseline shift issue, min.#points to detect baseline shift (default = 3, analysis skipped if set to 0).
USHP	For u-shape curves, min.#points to avoid flattening (default = 4, analysis skipped if set to 0).
TrustHi	For equal sets of corrections, trusts those retaining measurements at high concentrations (default = FALSE).
StrictImp	It prevents extrapolating over concentration-range boundaries; used for POD, ECxx etc (default = TRUE).
DUMV	A dummy value, default = -999.
TLOG	A scaling factor for calculating the wAUC, default = -24.
...	allow other parameters to pass

Value

A list with corrected concentration-response measurements and several calculated curve metrics.

• resp: corrected responses

• corr: flags for corrections

• ECxx: effective concentration values at various thresholds

• Cxx: concentrations for various absolute response levels

• Emax: maximum effective concentration, slope of the mid-curve (b/w EC25 and EC75)

• wConc: response-weighted concentration

• wResp: concentration-weighed response

• POD: point-of-departure (first concentration with response >TRSH)

• AUC: area-under-curve (in units of log-concentration X response)

• wAUC: AUC weighted by concentration range and POD / TLOG (-24)

• wAUC_pre: AUC weighted by concentration range and POD

• nCorrected: number of points corrected (basically, sum of flags in corr)

• Comments: warning and notes about the dose-response curve

• Settings: input parameters for this run

References

Sedykh A, Zhu H, Tang H, Zhang L, Richard A, Rusyn I, Tropsha A (2011). “Use of in vitro HTS-derived concentration-response data as biological descriptors improves the accuracy of QSAR models of in vivo toxicity.” Environmental health perspectives, 119(3), 364-370. ISSN 0091-6765, doi:10.1289/ehp.1002476.

Sedykh A (2016). “CurveP Method for Rendering High-Throughput Screening Dose-Response Data into Digital Fingerprints.” Methods in molecular biology (Clifton, N.J.), 1473. ISSN 1064-3745, doi:10.1007/978-1-4939-6346-1_14.

See Also

run_rcurvep() and combi_run_rcurvep()

Examples

curvep(Conc = c(-8, -7, -6, -5, -4) , Resp = c(0, -3, -5, -15, -30))

---

Default parameters of Curvep

Description

Default parameters of Curvep

Usage

curvep_defaults()

Value

A list of parameters with class as curvep_config.

• TRSH: (default = 15) base(zero-)line threshold

• RNGE: (default = -1000000, decreasing) target range of responses

• MXDV: (default = 5) maximum allowed deviation from monotonicity

• CARR: (default = 0) carryover detection threshold (analysis skipped if set to 0)

• BSFT: (default = 3) for baseline shift issue, min.#points to detect baseline shift (analysis skipped if set to 0)

• USHP: (default = 4) for u-shape curves, min.#points to avoid flattening (analysis skipped if set to 0)

• TrustHi: (default = TRUE)for equal sets of corrections, trusts those retaining measurements at high concentrations

• StrictImp: (default = TRUE) prevents extrapolating over concentration-range boundaries; used for POD, ECxx etc.

• DUMV: (default = -999) dummy value for inactive (not suggested to modify)

• TLOG: (default = -24) denominator for calculation wAUC (not suggested to modify)

• seed: (default = NA) can be set when bootstrapping samples

See Also

curvep()

Examples

# display all default settings
curvep_defaults()

# customize settings
custom_settings <- curvep_defaults()
custom_settings$TRSH <- 30
custom_settings

---

Estimate benchmark response (BMR) for each dataset

Description

Currently two methods have been implemented to get the "keen-point" from the variance(y) - threshold(x) curve. One is to use the original y values to draw a straight line between the lowest x value (p1) to highest x value (p2). The knee-point is the x that has the longest distance to the line. The other one is to fit the data first then use the fitted responses to do the same analysis. Currently the first method is preferred.

Usage

estimate_dataset_bmr(d, p1 = NULL, p2 = NULL, plot = TRUE)

Arguments

d	The rcurvep object with multiple samples and TRSHs. See combi_run_rcurvep() for an example.
p1	Default = NULL, or an integer value to manually set the first index of line.
p2	Default = NULL, or an integer value to manually set the last index of line.
plot	Default = TRUE, plot the diagnostic plot.

Details

The estimated BMR can be used in the calculation of POD. For example, if bmr = 25. For Curvep, combi_run_rcurvep(zfishbeh, TRSH = 25).
For Hill fit, summarize_fit_output(run_fit(zfishbeh, modls = "hill"), thr_resp = 25, extract_only = TRUE).

Value

A list with two components: stats and outcome.

• stats: a tibble, including pooled variance (pvar), fitted responses (y_exp_fit, y_lm_fit), distance to the line (dist2l)

• outcome: a tibble, including estimated BMRs (bmr)

; Suffix in the stats and outcome tibble: "ori" (original values), "exp"(exponential fit). prefix in the outcome tibble, "cor" (correlation between the fitted responses and the original responses), "bmr" (benchmark response), "qc" (quality control).

See Also

cal_knee_point(), combi_run_rcurvep()

Examples

# no extra cleaning
data(zfishdev_act)
bmr_out <- estimate_dataset_bmr(zfishdev_act, plot = FALSE)
plot(bmr_out)

# if want to do extra cleaning...
actm <- summarize_rcurvep_output(zfishdev_act, clean_only = TRUE, inactivate = "CARRY_OVER")

bmr_out <- estimate_dataset_bmr(actm, plot = FALSE)

---

Fit concentration-response data using Curve Class2 approach

Description

Curve Class2 uses 4-parameter Hill model to fit the data. The algorithm assumes the responses are in percentile. Curve Class2 classifies the curves based on fit quality and response magnitude.

Usage

fit_cc2_modl(Conc, Resp, classSD = 5, minYrange = 20, ...)

Arguments

Conc	A vector of log10 concentrations.
Resp	A vector of numeric responses.
classSD	A standard deviation (SD) derived from the responses in the vehicle control. it is used for classification of the curves. Default = 5%.
minYrange	A minimum response range (max activity - min activity) required to apply curve fitting. Curve fitting will not be attempted if the response range is less than the cutoff. Default = 20%.
...	for additional curve class2 parameters (currently none)

Details

cc2 = 1.1

2-asymptote curve, pvalue < 0.05, emax > 6\*classSD

cc2 = 1.2

2-asymptote curve, pvalue < 0.05, emax <= 6\*classSD & emax > 3\*classSD

cc2 = 1.3

2-asymptote curve, pvalue >= 0.05, emax > 6\*classSD

cc2 = 1.4

2-asymptote curve, pvalue >= 0.05, emax <= 6\*classSD & emax > 3\*classSD

cc2 = 2.1

1-asymptote curve, pvalue < 0.05, emax > 6\*classSD

cc2 = 2.2

1-asymptote curve, pvalue < 0.05, emax <= 6\*classSD & emax > 3\*classSD

cc2 = 2.3

1-asymptote curve, pvalue >= 0.05, emax > 6\*classSD

cc2 = 2.4

1-asymptote curve, pvalue >= 0.05, emax <= 6\*classSD & emax > 3\*classSD

cc2 = 3

single point activity, pvalue = NA, emax > 3\*classSD

cc2 = 4

inactive, pvalue >= 0.05, emax <= 3\*classSD

cc2 = 5

inconclusive, high bt, further investigation is needed

Value

A list of output parameters from Curve Class2 model fit. If the data are fit or not fittable (fit = 0), the default value for tp, ga, gw, bt pvalue, masks, nmasks is NA. For cc2 = 4, it is still possible to have fit parameters.

• modl: model type, i.e., cc2

• fit: fittable, 1 (yes) or 0 (no)

• aic: NA, it is not calculated for this model. The parameter is kept for compatability.

• cc2: curve class2, default = 4

• tp: model top, <0 means the fit for decreasing direction is preferred

• ga: ac50 (log10 scale)

• gw: Hill coefficient

• bt: model bottom

• pvalue: from F-test, for fit quality

• r2: fitness

• masks: a string to indicate at which positions of response are masked

• nmasks: number of masked responses

References

Huang R (2022). “A Quantitative High-Throughput Screening Data Analysis Pipeline for Activity Profiling.” Methods in molecular biology (Clifton, N.J.), 2474, 133—145. ISSN 1064-3745, doi:10.1007/978-1-0716-2213-1_13.

See Also

fit_modls()

Examples

fit_cc2_modl(c(-9, -8, -7, -6, -5, -4), c(0, 2, 30, 40, 50, 60))

---

Fit one set of concentration-response data using types of models

Description

A convenient function to fit data using available models and to sort the outcomes by AIC values.

Usage

fit_modls(Conc, Resp, Mask = NULL, modls, ...)

Arguments

Conc	A vector of log10 concentrations.
Resp	A vector of numeric responses.
Mask	Default = NULL or a vector of 1 or 0. 1 is for masking the respective response.
modls	The model types for the fitting. Currently available models are 3-parameter Hill model (hill), constant model (cnst), and Curve Class2 4-parameter Hill model (cc2). Multiple values are only allowed for the hill and cnst combination.
...	The named input configurations for replacing the default configurations. The input configuration needs to add model type as the prefix. For example, hill_pdir = -1 will set the Hill fit only to the decreasing direction. Another common parameter for cc2 model is cc2_classSD. The default value of cc2_classSD is 5%, which might be too small for noiser endpoints.

Details

The backbone of fit method using hill (3-parameter Hill model) and cnst (constant model) is based on the implementation from tcpl package. But the lower bound of ga is lower by log10(1/100). The cc2 model is the 4-parameter Hill model from Curve Class2.

Value

A list of components named by the models. The models are sorted by their AIC values (when multiple models are used). Thus, the first component has the best fit.

hill

Fit output from Hill equation

• modl: model type, i.e., hill

• fit: fittable, 1 (yes) or 0 (no)

• aic: AIC value

• tp: model top, <0 means the fit for decreasing direction is preferred

• ga: ac50 (log10 scale)

• gw: Hill coefficient

• er: scale term for Student's t distribution

cnst

Fit output from constant model

• modl: model type, i.e., cnst

• fit: fittable?, 1 or 0

• aic: AIC value

• er: scale term

cc2

Fit output from Curve Class 2 model

• modl: model type, i.e., cc2

• fit: fittable, 1 (yes) or 0 (no)

• aic: NA, it is not calculated for this model. The parameter is kept for compatability.

• cc2: curve class2, default = 4

• tp: model top, <0 means the fit for decreasing direction is preferred

• ga: ac50 (log10 scale)

• gw: Hill coefficient

• bt: model bottom

• pvalue: from F-test, for fit quality

• r2: fitness

• masks: a string to indicate at which positions of response are masked

• nmasks: number of masked responses

See Also

tcpl::tcplObjHill(), tcpl::tcplObjCnst(), get_hill_fit_config() fit_cc2_modl()

Examples

concd <- c(-9, -8, -7, -6, -5, -4)
respd <- c(0, 2, 30, 40, 50, 20)
maskd <- c(0, 0, 0, 0, 0, 1)

# run hill only
fit_modls(concd, respd, modls = "hill")

# run hill only + increasing direction only
fit_modls(concd, respd, modls = "hill", hill_pdir = 1)

# run cc2 only + change of classSD
fit_modls(concd, respd, modls = "cc2", cc2_classSD = 10)

# run hill + cnst
fit_modls(concd, respd, modls = c("hill", "cnst"))

# run with mask at the highest concentration
fit_modls(concd, respd, maskd, modls = "hill")

---

Get the default configurations for the Hill fit

Description

The function gives the default settings by using one set of concentration-response data.

Usage

get_hill_fit_config(Conc, Resp, optimf = "tcplObjHill")

Arguments

Conc	A vector of log10 concentrations.
Resp	A vector of numeric responses.
optimf	The default optimized function is tcpl::tcplObjHill(). but can be changed to ObjHillnorm().

Value

A list of input configurations.

• theta: initial values of parameters for Hill equation: tp, ga, gw, er

• f: the object function

• ui: the bound matrix

• ci: the bound constraints

See Also

tcpl::tcplObjHill(), fit_modls()

---

Merge results from multiple rcurvep objects

Description

Sometimes user may want to try multiple curvep setting and pick the one that can capture the shape (wAUC != 0). The highest absolute wAUC from the chemical-endpoint(-sample_id) pair will be picked.

Usage

merge_rcurvep_objs(...)

Arguments

...	rcurvep objects

Value

an updated rcurvep object with config = NULL

Examples

data(zfishbeh)

# combine default + mask
out1 <- combi_run_rcurvep(zfishbeh, TRSH = 10)
out2 <- combi_run_rcurvep(zfishbeh, TRSH = 10, mask = 1)
m1 <- merge_rcurvep_objs(out1, out2)

# use same set of samples to combine
out1 <- combi_run_rcurvep(zfishbeh, TRSH = 10, n_samples = 2, seed = 300)
out2 <- combi_run_rcurvep(zfishbeh, TRSH = 10, mask = 1, n_samples = 2, seed = 300)
m1 <- merge_rcurvep_objs(out1, out2)

---

Plot BMR diagnostic curves

Description

Plot BMR diagnostic curves

Usage

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

Arguments

x	The rcurvep_bmr object from estimate_dataset_bmr().
...	Allowed values: n_in_page, number of endpoints in a page.

Value

A ggplot object.

Examples

data(zfishdev_act)
bmr_out <- estimate_dataset_bmr(zfishdev_act, plot = FALSE)
plot(bmr_out)

---

Run parametric fits using types of models on concentration-response datasets

Description

Confidence intervals of activity metrics can be obtained through bootstrap approach. The bootstrap samples are generated by adding the residuals (the difference between the original responses and the Hill fit) to the fitted response (only for Hill equation, 3-parameter).

Usage

run_fit(d, modls, keep_sets = c("fit_set", "resp_set"), n_samples = NULL, ...)

Arguments

d	Datasets with concentration-response data. An example is zfishbeh. mask column is optional.
modls	The model types for the fitting. Currently available models are 3-parameter Hill model (hill), constant model (cnst), and Curve Class2 4-parameter Hill model (cc2). Multiple values are only allowed for the hill and cnst combination.
keep_sets	Output datasets. Multiple values are allowed. Default values are fit_set and resp_set. fit_set is a must. fit_set: a tibble with output from model fits resp_set: a tibble with fitted response data from the winning model. If winning model is hill + no fit or cc2 + hit=4(inactive), response is NA. If winning model is cnst, median of all responses is reported for each concentration.
n_samples	NULL (default) for no bootstrap samples are generated or number of samples to be generated from bootstrapping. When n_samples is not NULL, modls currently needs to be hill.
...	The named input configurations for replacing the default configurations. The input configuration needs to add model type as the prefix. For example, hill_pdir = -1 will set the Hill fit only to the decreasing direction. Add cc2_classSD = 10 will set the classification SD to 10%. Often 5% or 10% are used.

Value

A list of named components: result and result_nested. The result component is also a list of output sets depending on the parameter, keep_sets. The result_nested component is a tibble with input data nested in a column, input, and output data nested in a column, output.

Data structure

output |- result (list) | |- fit_set | |- resp_set | |- result_nested (tibble)

The prefix of the column names in the fit_set are the used models. The win_modl is the winning model.

See Also

fit_modls() for model fit information and the following analyses using summarize_fit_output(). for dichotomous response (see zfishdev), use create_dataset() first.

Examples

# It is suggested to use na.omit on the dataset to see if any data will be removed

# use hill + cnst model
fitd <- run_fit(zfishbeh, modls = c("hill", "cnst"))

# use only hill model and fit only to the decreasing direction, keep only the fit_set output
fitd <- run_fit(zfishbeh, modls = "hill", keep_sets = "fit_set", hill_pdir = -1)

# use cc2 model + higher classification SD
fitd <- run_fit(zfishbeh, modls = "cc2", cc2_classSD = 10)

# fit to the bootstrap samples using hill
fitd <- run_fit(zfishbeh, n_samples = 2, modls = "hill")

---

Run Curvep on datasets of concentration-response data

Description

The concentration-response relationship per endpoint and chemical has to be 1-to-1. If not, use create_dataset() for pre-processing or use combi_run_rcurvep(), which has both pre-processing and more flexible parameter controls.

Usage

run_rcurvep(
  d,
  mask = 0,
  config = curvep_defaults(),
  keep_sets = c("act_set", "resp_set", "fp_set"),
  ...
)

Arguments

d	Datasets with columns: endpoint, chemical, conc, and resp, mask (optional) Example datasets as zfishbeh. It is required that the baseline of responses in the resp column to be 0.
mask	Default = 0, for no mask (values in the mask column all 0). Use a vector of integers to mask the responses: 1 to mask the response at the highest concentration; 2 to mask the response at the second highest concentration, and so on. If mask column exists, the setting will be ignored.
config	Default configurations set by curvep_defaults().
keep_sets	The types of output to be reported. Allowed values: act_set, resp_set, fp_set. Multiple values are allowed. act_set is the must. act_set: activity data resp_set: response data fp_set: fingerprint data
...	Curvep settings. See curvep_defaults() for allowed parameters. These can be used to overwrite the default values.

Value

An rcurvep object. It has two components: result, config The result component is also a list of output sets depending on the parameter, keep_sets. The config component is a curvep_config object.

Often used columns in the act_set: AUC (area under the curve), wAUC (weighted AUC), POD (point-of-departure), EC50 (Half maximal effective concentration), nCorrected (number of corrected points).

See Also

create_dataset(), combi_run_rcurvep(), curvep_defaults().

Examples

data(zfishbeh)
d <- create_dataset(zfishbeh)

# default
out <- run_rcurvep(d)

# change TRSH
out <- run_rcurvep(d, TRSH = 30)

# mask response at highest and second highest concentration
out <- run_rcurvep(d, mask = c(1, 2))

---

Summarize the results from the parametric fitting using types of models

Description

The function first extracts the activity data based on the fit the supplied input parameters. In addition, summary of activity data (e.g., confidence interval, hit confidence) can be produced.

Usage

summarize_fit_output(
  d,
  thr_resp = 20,
  perc_resp = 10,
  ci_level = 0.95,
  extract_only = FALSE
)

Arguments

d	The output from the run_fit().
thr_resp	The response cutoff to calculate the potency. Default = 20 (POD20)
perc_resp	The percentage cutoff to calculate the potency. Default = 10 (EC10).
ci_level	The confidence level for the activity metrics. Default is = 0.95.
extract_only	Whether act_summary data should be produced. Default = FALSE.

Details

A tibble, act_set is generated. When (extract_only = FALSE), a tibble, act_summary is generated with confidence intervals of the activity metrics. The quantile approach is used to calculate the confidence interval. Currently only bootstrap calculations from hill (3-parameter) can generate confidence interval For potency activity metrics, if value is NA, highest tested concentration is used in the summary. For other activity metrics, if value is NA, 0 is used in the summary.

Value

A list of named components: result and result_nested (and act_summary). The result and result_nested are the copy from the output of run_fit(). An act_set is added under the result component. If (extract_only = FALSE), an act_summary is added.

Hit definition

cnst

If the cnst is the winning model and the median of responses larger than the thr_resp, it is considered as an hit. The median of responses is reported as Emax and the lowest tested concentration is reported as EC50, POD, ECxx.

hill

The hit (=1) is considered having POD < max tested concentration.

cc2

The hit value is from the cc2 value

Output structure

output |- result (list) | |- fit_set (tibble, all output from the respective fit model included) | |- resp_set (tibble) | |- act_set (tibble, EC50, ECxx, Emax, POD, slope, hit) | |- result_nested (tibble) |- act_summary (tibble, confidence interval)

activity metrics

hit

hit call, see above definition

EC50

half maximal effect concentration

ECxx

effect concentration at XX percent, depending on the perc_resp

POD

point-of-departure, depending on the thr_resp

Emax

max effect - min effect from the fit

slope

slope factor from the fit

See Also

run_fit()

Examples

# generate some fit outputs


## fit only
fitd1 <- run_fit(zfishbeh, modls = "cc2")

## fit + bootstrap samples
fitd2 <- run_fit(zfishbeh, n_samples = 3, modls = "hill")

## fit using hill + cnst
fitd3 <- run_fit(zfishbeh, modls = c("hill", "cnst"))


# only to extract the activity data
sumd1 <- summarize_fit_output(fitd1, extract_only = TRUE)
sumd3 <- summarize_fit_output(fitd3, extract_only = TRUE)

# calculate EC20 instead of default EC10
sumd1 <- summarize_fit_output(fitd1, extract_only = TRUE, perc_resp  = 20)

# calculate POD using a higher noise level (e.g., 40)
## this number depends on the response unit
sumd1 <- summarize_fit_output(fitd1, extract_only = TRUE, thr_resp  = 40)

# calculate confidence intervals based on the bootstrap samples
sumd2 <- summarize_fit_output(fitd2)

---

Clean and summarize the output of rcurvep object

Description

Clean and summarize the output of rcurvep object

Usage

summarize_rcurvep_output(
  d,
  inactivate = NULL,
  ci_level = 0.95,
  clean_only = FALSE
)

Arguments

d	The rcurvep object from combi_run_rcurvep() and run_rcurvep().
inactivate	A character string, default = NULL, to make the curve with this string in the Comments column as inactive. or a vector of index for the rows in the act_set that needs to be inactive
ci_level	Default = 0.95 (95 percent of confidence interval).
clean_only	Default = FALSE, only the 1st, 2nd task will be performed (see Details).

Details

The function can perform the following tasks:

1. add an column, hit, in the act_set

2. unhit (make result as inactive) if the Comments column contains a certain string

3. summarize the results

The curve is considered as "hit" if its responses are monotonic after processing by Curvep. However, often, if the curve is "INVERSE" (yet monotonic) is not considered as an active curve. By using the information in the Comments column, we can "unhit" these cases.

When (clean_only = FALSE, default), a tibble, act_summary is generated with confidence intervals of the activity metrics. The quantile approach is used to calculate the confidence interval. For potency activity metrics, if value is NA, highest tested concentration is used in the summary. For other activity metrics, if value is NA, 0 is used in the summary.

Value

A list of named components: result and config (and act_summary). The result and config are the copy of the input d (but with modifications if inactivate is not NULL). If (clean_only = FALSE), an act_summary is added.

Suffix meaning in column names in act_summary: med (median), cil (lower end confidence interval), ciu (higher end confidence interval) Often used columns in act_summary: n_curves (number of curves used in summary), hit_confidence (fraction of active in n_curves)

See Also

combi_run_rcurvep(), run_rcurvep()

Examples

data(zfishbeh)

# original datasets
out <- combi_run_rcurvep(zfishbeh, n_samples = NULL, TRSH = c(5, 10))
out_res <- summarize_rcurvep_output(out)


# unhit when comment has "INVERSE"
out <- summarize_rcurvep_output(out, inactivate = "INVERSE")

# unhit for certain rows in act_set
out <- summarize_rcurvep_output(out, inactivate = c(2,3))

# simulated datasets
out <- combi_run_rcurvep(zfishbeh, n_samples = 3, TRSH = c(5, 10))
out_res <- summarize_rcurvep_output(out)

---

Subsets of concentration response datasets from zebrafish neurotoxicity assays

Description

The datasets contain 11 toxicity endpoints and 2 chemicals. The responses have been normalized so that the baseline is 0.

Usage

zfishbeh

Format

A tibble with 2123 rows and 4 columns:

endpoint

endpoint name

chemical

chemical name + CASRN

conc

concentrations in log10(M) format

resp

responses after normalized using the vehicle control on each plate

Source

Biobide study S-BBD-0017/15

---

Subsets of concentration response datasets from zebrafish developmental toxicity assays

Description

The datasets contain 4 toxicity endpoints and 3 chemicals.

Usage

zfishdev

Format

A tibble with 96 rows and 5 columns:

endpoint

endpoint name + at time point measured

chemical

chemical name + CASRN

conc

concentrations in log10(M) format

n_in

number of incidence

N

number of embryos

Source

Biobide study S-BBD-00016/15

---

Activity output based on simulated datasets using zfishdev_all dataset

Description

The data is an rcurvep object from the combi_run_rcurvep(). See combi_run_rcurvep() for the code to reproduce this dataset.

Usage

zfishdev_act

Format

A list of two named components: result and config. The result component is a list with one component: act_set.

See Also

estimate_dataset_bmr()

---

Full sets of concentration response datasets from zebrafish developmental toxicity assays

Description

The datasets contain 4 toxicity endpoints and 32 chemicals.

Usage

zfishdev_all

Format

A tibble with 512 rows and 5 columns:

Source

Biobide study S-BBD-00016/15

See Also

zfishdev

---