Package 'profrep'

Description

Calculate the Number of Crossovers

Usage

calculate_crossovers(individual_df, n_trials, n_replicates)

Arguments

Details

This function calculates the number of crossovers in a dataset by comparing the values of replicates across multiple trials. It assumes that missing values (NAs) have been interpolated using the clean_data function.

Value

The number of crossovers detected in the dataset.

See Also

clean_data for information on data cleaning.

Examples

data <- matrix(
   c(
     1, 60, 1, 2, 3, 4, 5,   # No NA values
     1, 90, 9, NA, 4, NA, 2,  # NA Values in row
     1, 120, 3, 6, NA, NA, 9  # Consecutive NA values
    ),
    nrow = 3,
    byrow=TRUE
)
n_trials <- nrow(data)
n_replicates <- ncol(data) - 2
crossovers <- calculate_crossovers(data, n_trials, n_replicates)
cat("Number of crossovers:", crossovers, "\n")

Description

Clean Data by Interpolating Missing Values

Usage

clean_data(data, n_trials, n_replicates)

Arguments

Details

This function cleans a dataset by interpolating missing values in the replicate columns of each row using neighboring values. If the data frame ends in null values (the last columns are nulls), it will extrapolate from the last value. If the first value is null, it will loop around and pull from the last replicate to perform the interpolation between the last replicate and the second replicate.

Value

A cleaned data frame with missing values interpolated.

See Also

find_next_good_datapoint for details on the interpolation process.

Examples

my_data <- matrix(
   c(
     1, 60, 1, 2, 3, 4, 5,   # No NA values
     1, 90, 9, NA, 4, NA, 2,  # NA Values in row
     1, 120, 3, 6, NA, NA, 9  # Consecutive NA values
    ),
    nrow = 3,
    byrow=TRUE
 )
cleaned_data <- clean_data(my_data, n_trials = 3, n_replicates = 5)
print(my_data)
print(cleaned_data)

Description

Score and Order Data

Usage

do_ordering(
  n_trials,
  id_list,
  df_list,
  n_replicates,
  verbose = FALSE,
  sort = TRUE
)

Arguments

Details

Performs the ordering of input data by scoring each individual data frame.

The main function of the package, this will send each individuals data out for scoring. Then, when all scores are computed, it will order the result data frame by score and assign a rank.

Ranks are assigned with ties allowed - if N individuals have a tie, their rank is averaged. For example, if the max score is 1, and two individuals have that score, their rank is 1.5

Value

Returns a data frame of the results, in the following form:

     - Column 1: "individual" - the unique identifier of an individual or sample
     - Column 2: "n_crossings" - the calculated number of crossings.
     - Column 3: "max_variance" - the maximum of the variances of the replicate measurements at a single time for the individual or sample.
     - Column 4: "ave_variance" - the average of the variances of the replicate measurements at a single time for the individual or sample.
     - Column 5: "base_score" - the original, unnormalized profile repeatability score. Smaller numbers rank higher.
     - Column 6: "final_score" - the base score, normalized by the sigmoid function. Constrained to be between 0 and 1. Scores closer to 1 rank higher.
     - Column 7: "rank" - the calculated ranking of the individual or sample, against all other individuals or samples in the data set.

Examples

df <- data.frame(
    col_a = c('A', 'A', 'B', 'B'),
    col_b = c(5, 15, 5, 15),
    col_c = c(5, 10, 1, 2),
    col_d = c(10, 15, 3, 4)
  )
id_list <- unique(df[, 1])
individuals <- list()
for (i in 1:length(id_list)) {
  individuals[[i]] <- df[df[, 1] == id_list[i], ]
}
ret_df <- do_ordering(n_trials=2, id_list=id_list, df_list=individuals, n_replicates=2)
print(ret_df)

Description

An example of data that one would perform profile repeatability on. Consists of 9 individual animals, with corticosterone data taken at 2 timepoints (n_trials = 2), baseline (time = 3) and stress-induced (time = 30). Then, there are 28 replicate columns.

Usage

example_two_point_data

Format

example_two_point_data

A dataframe with 10 rows and 30 columns:

Animal

The animal name/unique identifier

Time

The time of the measurement, in days.

SD.DR

The name of the replicate column.

Source

This data was extracted from Romero & Rich 2007 (Comp Biochem. Physiol. Part A Mol. Integr. Physiol. 147, 562-568. https://doi.org/10.1016/j.cbpa.2007.02.004)

Description

What Is the Next Non-Null Data Point?

Usage

find_next_good_datapoint(data_row, index, n_replicates)

Arguments

Details

Given a data row, an index, and the number of replicates (the number of elements in the row), this function finds the next good data point in the row.

A good data point is a non-missing value (not NA) with a non-empty string.

Value

The next good data point or -999 if none is found.

Examples

data_row <- c(NA, 3, 2, NA, 5)
index <- 1
n_replicates <- 5
find_next_good_datapoint(data_row, index, n_replicates) # expect 3

Description

Calculate Group Variance

Usage

get_vars(individual_array, n_replicates)

Arguments

Details

For the given individual array, for all rows of times, computes the variance in values over replicates.

Returns these variances, sum of all values (for all times and replicates), sum of all these values squared, and the number of values.

Value

A list, where the elements are:

     1. variances: A vector of the variances of the sample
     2. total_sum: The sum of all the measurements in the sample
     3. ssq: The sum of all the squares of the measurements in the sample
     4. num_measurements: The total number of measurements in the sample that are not null

Examples

arr <- data.frame(
  individual=c("a", "a"),
  time=c(5, 15),
  col_a=c(1, 2),
  col_b=c(2, 3)
)
variance_return <- get_vars(individual_array=arr, n_replicates=2)
print(variance_return)

Description

Perform Profile Repeatability

Usage

profrep(df, n_timepoints, sort = TRUE, verbose = FALSE)

Arguments

            | id | time | A | B |
            |:--:|:----:|:-:|:-:|
            | 1  | 15 | 1 | 2 |
            | 1  | 30 | 3 | 4 |

Details

Calculates the profile repeatability measure of the input data according to the method in Reed et al., 2019, J. Gen. Comp. Endocrinol. (270).

Value

Returns a data frame of the results, in the following form:

• Column 1: "individual" - the unique identifier of an individual or sample

• Column 2: "n_crossings" - the calculated number of crossings.

• Column 3: "max_variance" - the maximum of the variances of the replicate measurements at a single time for the individual or sample.

• Column 4: "ave_variance" - the average of the variances of the replicate measurements at a single time for the individual or sample.

• Column 5: "base_score" - the original, unnormalized profile repeatability score. Smaller numbers rank higher.

• Column 6: "final_score" - the base score, normalized by the sigmoid function. Constrained to be between 0 and 1. Scores closer to 1 rank higher.

• Column 7: "rank" - the calculated ranking of the individual or sample, against all other individuals or samples in the data set.

See Also

do_ordering for the main data processing function.

calculate_crossovers for how the number of crossings are calculated.

score_individual_df for how the score is calculated for an individual or sample.

clean_data for how missing replicate values are handled.

Examples

test_data <- profrep::example_two_point_data
results <- profrep::profrep(df=test_data, n_timepoints=2)
print(results)

Description

This function retrieves the indices of non-missing data values at a specific time point from an individual array.

Usage

retrieve_good_data(individual_array, t, n_replicates)

Arguments

Value

A numeric vector containing the indices of non-missing data values at the specified time point t. If there are no non-missing values or only one non-missing value, NULL is returned.

See Also

which function for finding the indices of non-missing values.

Examples

# Example usage:
individual_data <- matrix(c(NA, 2, NA, 4, 5, NA), nrow = 1)
retrieve_good_data(individual_data, t = 1, n_replicates = 3)

Description

Compute Profile Repeatability Score

Usage

score_dfs(id_list, df_list, n_replicates, n_trials, verbose = FALSE)

Arguments

Details

Works on multiple elements of data.

Splits the data into the data frame for a particular individual from the id_list, then calculates metrics to compute the profile repeatability score. Returns a data frame with the individuals name and the score.

Value

A dataframe of the calculated metrics. The column structure is as follows:

     - Column 1: "individual" - the unique identifier of an individual or sample
     - Column 2: "n_crossings" - the calculated number of crossings.
     - Column 3: "max_variance" - the maximum of the variances of the replicate measurements at a single time for the individual or sample.
     - Column 4: "ave_variance" - the average of the variances of the replicate measurements at a single time for the individual or sample.
     - Column 5: "base_score" - the original, unnormalized profile repeatability score. Smaller numbers rank higher.
     - Column 6: "final_score" - the base score, normalized by the sigmoid function. Constrained to be between 0 and 1. Scores closer to 1 rank higher.

Examples

df <- data.frame(
    col_a = c('A', 'A', 'B', 'B'),
    col_b = c(5, 15, 5, 15),
    col_c = c(5, 10, 1, 2),
    col_d = c(10, 15, 3, 4)
  )
id_list <- unique(df[, 1])
individuals <- list()
for (i in 1:length(id_list)) {
  individuals[[i]] <- df[df[, 1] == id_list[i], ]
}
ret_df <- score_dfs(id_list=id_list, df_list=individuals, n_replicates=2, n_trials=2)
print(ret_df)

Description

Score an Individual Data Frame

Usage

score_individual_df(
  individual_df,
  n_trials,
  n_replicates,
  max_variance,
  variance_set
)

Arguments

Details

This function calculates a score for an individual data frame based on various factors, including the number of crossovers, maximum variance, and a set of variances.

The score is computed as follows:

• It factors in the number of crossovers using a scaling factor.

• It considers the maximum variance value in the variance set.

• It adds a component based on the average of variance values.

• It includes a scaled component of the number of crossovers.

Value

A list calculated for the individual data frame. Contains two values:

1. n_crossings: The number of crossover events in the data.

2. base_score: The un-normalized profile repeatability score for the data.

See Also

calculate_crossovers for information on crossovers calculation.

Examples

arr <- data.frame(
  individual=c("a", "a"),
  time=c(5, 15),
  col_a=c(1, 2),
  col_b=c(2, 3)
)
variance_set <- c(0.5, 0.5)
max_variance <- 0.5
score_list <- score_individual_df(
  individual_df=arr,
  n_trials=2,
  n_replicates=2,
  max_variance=max_variance,
  variance_set=variance_set
)
print(score_list)

Description

Calculates the sigmoid function of the input

Usage

sigmoid(float)

Arguments

Value

A float number which is the result of the sigmoid function

Examples

sigmoid(0)
sigmoid(2)

Description

An example of data that one would perform profile repeatability on. Consists of 12 individual animals, with corticosterone data taken at 3 times (n_trials = 3), baseline (time = 0) and two stress-induced (time = 15 and 30). Then, there are 10 replicate columns. This example also shows what happens when there are null data records for some individuals.

Usage

sparrow_repeatability_three_point

Format

sparrow_repeatability_three_point

A dataframe with 36 rows and 12 columns:

Animal

The animal name/unique identifier

TIME

The time of the measurement, in days

LD.500

The name of the replicate column

Source

This data was extracted from Rich & Romero 2001 (J. Comp. Physiol. Part B Biochem. Syst. Environ. Physiol. 171, 543-647. https://doi.org/10.1007/s003600100204)

Description

An example of data that one would perform profile repeatability on. The data is synthetic data created for testing purposes and is designed to span a range of perceived repeatability scores. Consists of 11 individual animals, with data taken at 4 times (n_trials = 4), baseline (time = 0) and three stress-induced (time = 15, 30, and 45). Then, there are four replicate columns. Replicate column names refer to sample tests performed on the animal.

Usage

synthetic_data_four_point

Format

synthetic_data_four_point

A dataframe with 44 rows and 6 columns:

Animal

The animal name/unique identifier

TIME

The time of the measurement (unit not important)

A

The (unimportant) name of a replicate column.

B

The (unimportant) name of a replicate column.

C

The (unimportant) name of a replicate column.

D

The (unimportant) name of a replicate column.

Source

Data created for testing purposes by Reed et al., 2019 (Gen. Comp. Endocrinol. 270, 1-9. https://doi.org/10.1016/j.ygcen.2018.09.015)