| Type: | Package |
| Title: | Evolutionary Version of the Metropolis-Hastings Algorithm |
| Version: | 1.2 |
| Date: | 2025-01-31 |
| Author: | Jules Bangard 
[aut, cre] |
| Maintainer: | Jules Bangard <jules.bangard@etu.unistra.fr> |
| Description: | Provides computational methods for detecting adverse high-order drug interactions from individual case safety reports using statistical techniques, allowing the exploration of higher-order interactions among drug cocktails. |
| License: | GPL-3 |
| Imports: | Rcpp (≥ 1.0.7), ggplot2, dplyr, umap, dbscan, stats |
| LinkingTo: | Rcpp, RcppArmadillo |
| RoxygenNote: | 7.2.3 |
| LazyData: | true |
| Suggests: | knitr, rmarkdown, gridExtra |
| VignetteBuilder: | knitr |
| NeedsCompilation: | yes |
| Packaged: | 2025-02-26 12:36:14 UTC; julesbangard |
| Depends: | R (≥ 3.5.0) |
| Repository: | CRAN |
| Date/Publication: | 2025-02-27 17:00:02 UTC |
Evolutionary Version of the Metropolis-Hastings Algorithm
Description
Provides computational methods for detecting adverse high-order drug interactions from individual case safety reports using statistical techniques, allowing the exploration of higher-order interactions among drug cocktails.
Author(s)
Jules Bangard [aut, cre] (<https://orcid.org/0009-0007-4670-7860>)
Maintainer: Jules Bangard <jules.bangard@etu.unistra.fr>
See Also
ATC Tree Upper Bound 2024
Description
Example dataset representing the ATC tree structure, sourced from the WHO website (2024-02-23). This dataset is provided for demonstration and testing purposes with the package.
Usage
ATC_Tree_UpperBound_2024
Format
A data frame with 4 variables:
- ATCCode
The code of ATC nodes
- Name
The name of ATC nodes
- ATC_length
The number of characters in the ATCCode
- upperBound
The index of the last child node in the tree
Source
World Health Organization, ATC classification register
Convert ATC Code for each patients to the corresponding DFS number of the ATC tree
Description
Convert ATC Code for each patients to the corresponding DFS number of the ATC tree
Usage
ATCtoNumeric(patientATC, tree)
Arguments
patientATC |
: patients observations, for each patient we got a string containing taken medications (ATC code) |
tree |
: ATC tree (we assume that there is a column 'ATCCode' ) |
Value
a matrix of the same size as patientATC but containing integer that are the index of the corresponding ATC code.
Examples
ATC_code <- c('A01AA30 A01AB03', 'A10AC30')
ATCtoNumeric(ATC_code, ATC_Tree_UpperBound_2024)
The MCMC method that runs the random walk on a single cocktail in order to estimate the distribution of score among cocktails of size Smax.
Description
The MCMC method that runs the random walk on a single cocktail in order to estimate the distribution of score among cocktails of size Smax.
Usage
DistributionApproximation(
epochs,
ATCtree,
observations,
temperature = 1L,
nbResults = 5L,
Smax = 2L,
p_type1 = 0.01,
beta = 4L,
max_score = 500L,
num_thread = 1L,
verbose = FALSE
)
Arguments
epochs |
: number of steps for the MCMC algorithm |
ATCtree |
: ATC tree with upper bound of the DFS (without the root, also see on the github repo for an example) |
observations |
: real observation of the AE based on the medications of each patients (a DataFrame containing the medication on the first column and the ADR (boolean) on the second) |
temperature |
: starting temperature, default = 1 (denoted T in the article) |
nbResults |
: Number of returned solution (Cocktail of size Smax with the best oberved score during the run), 5 by default |
Smax |
: Size of the cocktail we approximate the distribution from |
p_type1 |
: probability to operate type1 mutation. Note : the probability to operate the type 2 mutation is then 1 - P_type1. P_type1 must be in [0;1]. Default is .01 |
beta |
: filter the minimum number of patients that must have taken the cocktail for his risk to be taken into account in the DistributionScoreBeta default is 4 |
max_score |
: maximum number the score can take. Score greater than this one would be added to the distribution as the value max_score. Default is 500 |
num_thread |
: Number of thread to run in parallel if openMP is available, 1 by default |
verbose |
: Output summary (default is false) |
Value
I no problem, return a List containing : - ScoreDistribution : the distribution of the score as an array with each cells representing the number of risks = (index-1)/ 10 - Outstanding_score : An array of the score greater than max_score, - Best_cocktails : the nbResults bests cocktails encountered during the run. - Best_scores : Score corresponding to the bestCocktails. - FilteredDistribution : Distribution containing score for cocktails taken by at least beta patients. - Best_cocktails_beta : the nbResults bests cocktails taken by at least beta patients encountered during the run. - Best_scores_beta : Score corresponding to the bestCocktailsBeta. - cocktailSize : Smax parameter used during the run. ; Otherwise the list is empty
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
estimation = DistributionApproximation(epochs = 10, ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy)
FAERS Myopathy Dataset
Description
Example dataset representing drug intake and adverse event reports from FAERS. This dataset is provided to demonstrate the functionality of genetic and MCMC algorithms in the package.
Usage
FAERS_myopathy
Format
A data frame with 2 columns:
- patientATC
Drug intake for each patient as a vector of ATC tree indices
- patientADR
Indicates if the patient experienced myopathy as an adverse event
Source
Food & Drug Administration Event Reporting System (FAERS)
Genetic algorithm, trying to reach riskiest cocktails (the ones which maximize the fitness function, Hypergeometric score in our case)
Description
Genetic algorithm, trying to reach riskiest cocktails (the ones which maximize the fitness function, Hypergeometric score in our case)
Usage
GeneticAlgorithm(
epochs,
nbIndividuals,
ATCtree,
observations,
num_thread = 1L,
diversity = FALSE,
p_crossover = 0.8,
p_mutation = 0.01,
nbElite = 0L,
tournamentSize = 2L,
alpha = 1,
summary = TRUE
)
Arguments
epochs |
: number of step or the algorithm |
nbIndividuals |
: size of the population |
ATCtree |
: ATC tree with upper bound of the DFS (without the root) |
observations |
: real observation of the AE based on the medications of each patients (a DataFrame containing the medication on the first column and the ADR (boolean) on the second) |
num_thread |
: Number of thread to run in parallel if openMP is available, 1 by default |
diversity |
: enable the diversity mechanism of the algorithm (favor the diversity of cocktail in the population), default is false |
p_crossover |
: probability to operate a crossover on the crossover phase. Default is 80% |
p_mutation |
: probability to operate a mutation after the crossover phase. Default is 1% |
nbElite |
: number of best individual we keep from generation to generation. Default is 0 |
tournamentSize |
: size of the tournament (select the best individual between tournamentSize sampled individuals) |
alpha |
: when making a type 1 mutation you have (alpha / size of cocktail) chance to add a drug. |
summary |
: print the summary of population at each steps ? |
Value
If no problem, return a List : - meanFitnesses : The mean score of the population at each epochs of the algorithm. - BestFitnesses : The best score of the population at each epochs of the algorithm. - FinalPopulation : The final population of the algorithm when finished (medications and corresponding scores)
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
results = GeneticAlgorithm(epochs = 10, nbIndividuals = 10,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy)
Output the outstanding score (Outstanding_score) outputed by the MCMC algorithm in a special format
Description
Output the outstanding score (Outstanding_score) outputed by the MCMC algorithm in a special format
Usage
OutsandingScoreToDistribution(outstanding_score, max_score)
Arguments
outstanding_score |
: Outstanding_score outputed by MCMC algorithm to be converted to the ScoreDistribution format |
max_score |
: max_score parameter used during the MCMC algorithm |
Value
outstanding_score in a format compatible with MCMC algorithm output
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
DistributionApproximationResults = DistributionApproximation(epochs = 10,
ATCtree = ATC_Tree_UpperBound_2024, observations = FAERS_myopathy)
OutsandingScoreToDistribution(DistributionApproximationResults$Outstanding_score, max_score = 100)
Calculate the divergence between 2 distributions (the true Distribution and the learned one)
Description
Calculate the divergence between 2 distributions (the true Distribution and the learned one)
Usage
calculate_divergence(
empirical_distribution,
true_distribution,
method = "TV",
Filtered = FALSE
)
Arguments
empirical_distribution |
A numeric vector of values representing the empirical distribution (return value of DistributionAproximation function) |
true_distribution |
A numeric vector of values representing the true distribution computed by the trueDistributionSizeTwoCocktail function |
method |
A string, either "TV" or "KL" to respectively use the total variation distance or the Kullback-Leibler divergence. (default = "TV") |
Filtered |
Should we use the filtered distribution or the normal one |
Value
A numeric value representing the divergence of the 2 distributions
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
estimated_score_distribution = DistributionApproximation(epochs = 10,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy[1:100,], Smax =2)
true_score_distribution = trueDistributionSizeTwoCocktail(ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy[1:100,], beta = 4)
divergence <- calculate_divergence(empirical_distribution = estimated_score_distribution,
true_distribution = true_score_distribution)
Clustering of the solutions of the genetic algorithm using the hclust algorithm
Description
Clustering of the solutions of the genetic algorithm using the hclust algorithm
Usage
clustering_genetic_algorithm(
genetic_results,
ATCtree,
dist.normalize = TRUE,
umap_config = NULL
)
Arguments
genetic_results |
A list of cocktails in the form of integer vector |
ATCtree |
ATC tree with upper bound of the DFS |
dist.normalize |
Do we normalize the distance (so it belongs to [0;1]) |
umap_config |
The configuration to use in order to project the cocktails in a smaller space (umap::umap.defaults by default) |
Value
A dataframe containing UMAP 1/2 the two coordinates of each cocktails in the plane as well as the cluster number of each cocktails
Examples
data("ATC_Tree_UpperBound_2024")
results = GeneticAlgorithm(epochs = 10, nbIndividuals = 10,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy)
hclust_genetic_solution(genetic_results = results,
ATCtree = ATC_Tree_UpperBound_2024)
Function used in the reference article to compare diverse Disproportionality Analysis metrics
Description
Function used in the reference article to compare diverse Disproportionality Analysis metrics
Usage
computeMetrics_size2(CocktailList, ATCtree, observations, num_thread = 1L)
Arguments
CocktailList |
: A list of cocktails on which the Disproportionality analysis metrics should be computed |
ATCtree |
: ATC tree with upper bound of the DFS (without the root) |
observations |
: observation of the AE based on the medications of each patients (a DataFrame containing the medication on the first column and the ADR (boolean) on the second) on which we want to compute the risk distribution |
num_thread |
: Number of thread to run in parallel if openMP is available, 1 by default |
Value
Multiple DA metrics computed on CocktailList cocktails
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
cocktails = list(c(561, 904),
c(1902, 4585)) # only size 2 cocktails allowed for this function
scores_of_cocktails = computeMetrics_size2(CocktailList = cocktails,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy[1:100,])
Function used to compute the Relative Risk on a list of cocktails
Description
Function used to compute the Relative Risk on a list of cocktails
Usage
compute_RR_on_list(cocktails, ATCtree, observations, num_thread = 1L)
Arguments
cocktails |
: A list containing cocktails in the form of vector of integers (ATC index) |
ATCtree |
: ATC tree with upper bound of the DFS (without the root) |
observations |
: observation of the AE based on the medications of each patients (a DataFrame containing the medication on the first column and the ADR (boolean) on the second) on which we want to compute the risk distribution |
num_thread |
: Number of thread to run in parallel if openMP is available, 1 by default |
Value
RR score among "cocktails" parameters
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
cocktails = list(c(561, 904),
c(1902, 4585))
RR_of_cocktails = compute_RR_on_list(cocktails = cocktails,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy)
Function used to compute the Hypergeometric score on a list of cocktails
Description
Function used to compute the Hypergeometric score on a list of cocktails
Usage
compute_hypergeom_on_list(cocktails, ATCtree, observations, num_thread = 1L)
Arguments
cocktails |
: A list containing cocktails in the form of vector of integers (ATC index) |
ATCtree |
: ATC tree with upper bound of the DFS (without the root) |
observations |
: observation of the AE based on the medications of each patients (a DataFrame containing the medication on the first column and the ADR (boolean) on the second) on which we want to compute the risk distribution |
num_thread |
: Number of thread to run in parallel if openMP is available, 1 by default |
Value
Hypergeometric score among "cocktails" parameters
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
cocktails = list(c(561, 904),
c(1902, 4585))
Hypergeom_of_cocktails = compute_hypergeom_on_list(cocktails = cocktails,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy)
Function used to convert your genetic algorithm results that are stored into a .csv file to a Data structure that can be used by the clustering algorithm
Description
Function used to convert your genetic algorithm results that are stored into a .csv file to a Data structure that can be used by the clustering algorithm
Usage
csv_to_population(ATC_name, filename, sep = ";")
Arguments
ATC_name |
the ATC_name column of the ATC tree |
filename |
Name of the file where the results are located |
sep |
the separator to use when opening the csv file (';' by default) |
Value
An R List that can be used by other algorithms (e.g. clustering algorithm)
Examples
data("ATC_Tree_UpperBound_2024")
genetic_results = csv_to_population(ATC_Tree_UpperBound_2024$Name,
"path/to/output.csv")
Recover the square matrix of distance between cocktails where the index (i,j) of the matrix is the distance between cocktails i and j in an arbitrary cocktail list
Description
Recover the square matrix of distance between cocktails where the index (i,j) of the matrix is the distance between cocktails i and j in an arbitrary cocktail list
Usage
get_dissimilarity_from_cocktail_list(cocktails, ATCtree, normalization = TRUE)
Arguments
cocktails |
: A list of cocktails in the form of a vector of integer |
ATCtree |
: ATC tree with upper bound of the DFS (without the root) |
normalization |
: Do we keep the distance between cocktail in the range [0;1] ? |
Value
The square matrix of distances between cocktails
Examples
data("ATC_Tree_UpperBound_2024")
cocktails = list(c(561, 904),
c(1902, 4585)) # only size 2 cocktails allowed for this function
distance_matrix = get_dissimilarity_from_cocktail_list(cocktails = cocktails,
ATCtree = ATC_Tree_UpperBound_2024,
normalization = TRUE)
Recover the square matrix of distance between cocktails where the index (i,j) of the matrix is the distance between cocktails i and j in the genetic_results list.
Description
Recover the square matrix of distance between cocktails where the index (i,j) of the matrix is the distance between cocktails i and j in the genetic_results list.
Usage
get_dissimilarity_from_genetic_results(genetic_results, ATCtree, normalization)
Arguments
genetic_results |
the List returned by the genetic algorithm. |
ATCtree |
: ATC tree with upper bound of the DFS (without the root) |
normalization |
: Do we keep the distance between cocktail in the range [0;1] ? |
Value
The square matrix of distances between cocktails
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
genetic_results = GeneticAlgorithm(epochs = 10, nbIndividuals = 10,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy)
distance_matrix = get_dissimilarity_from_genetic_results(genetic_results = genetic_results,
ATCtree = ATC_Tree_UpperBound_2024, normalization = TRUE)
Recover the square matrix of distance between cocktails where the index (i,j) of the matrix is the distance between cocktails i and j in the csv file containing results of genetic algorithm
Description
Recover the square matrix of distance between cocktails where the index (i,j) of the matrix is the distance between cocktails i and j in the csv file containing results of genetic algorithm
Usage
get_dissimilarity_from_txt_file(filename, ATCtree, normalization = TRUE)
Arguments
filename |
: the name of the file returned by the print_csv function. |
ATCtree |
: ATC tree with upper bound of the DFS (without the root) |
normalization |
: Do we keep the distance between cocktail in the range [0;1] ? |
Value
The square matrix of distances between cocktails
Examples
data("ATC_Tree_UpperBound_2024")
distance_matrix = get_dissimilarity_from_txt_file(filename = '250e_700ind_0.2mr_0ne_2alpha.txt',
ATCtree = ATC_Tree_UpperBound_2024, normalization = TRUE)
Clustering of the solutions of the genetic algorithm using the hclust algorithm
Description
Clustering of the solutions of the genetic algorithm using the hclust algorithm
Usage
hclust_genetic_solution(
genetic_results,
ATCtree,
dist.normalize = TRUE,
method = "complete"
)
Arguments
genetic_results |
The return value of the genetic algorithm |
ATCtree |
ATC tree with upper bound of the DFS |
dist.normalize |
Do we normalize the distance (so it bellongs to [0;1]) |
method |
(from hclust function) the agglomeration method to be used. This should be (an unambiguous abbreviation of) one of "ward.D", "ward.D2", "single", "complete", "average" (= UPGMA), "mcquitty" (= WPGMA), "median" (= WPGMC) or "centroid" (= UPGMC). |
Value
the hierarchical clustering of the results of the genetic algorithm
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
results = GeneticAlgorithm(epochs = 10, nbIndividuals = 10,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy)
hclust_genetic_solution(genetic_results = results,
ATCtree = ATC_Tree_UpperBound_2024)
Convert the histogram returned by the DistributionApproximation function, to a real number distribution (that can be used in a test for example)
Description
Convert the histogram returned by the DistributionApproximation function, to a real number distribution (that can be used in a test for example)
Usage
histogramToDitribution(vec)
Arguments
vec |
: distribution returned by the DistributionAproximationFunction |
Value
A vector containing sampled risk during the MCMC algorithm
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
DistributionApproximationResults = DistributionApproximation(epochs = 10,
ATCtree = ATC_Tree_UpperBound_2024, observations = FAERS_myopathy)
histogramToDitribution(DistributionApproximationResults$ScoreDistribution)
This function can be used in order to try different set of parameters for the genetic algorithm in a convenient way. This will run each combination of mutation_rate, nb_elite and alphas possible nb_test_desired times. For each sets of parameters, results will be saved in a file named according to the set of parameter. One can regroup the results of each run in a csv file by using the print_csv function specifying the names of each file that needs to be treated and the number of performed runs on each parameter set
Description
This function can be used in order to try different set of parameters for the genetic algorithm in a convenient way. This will run each combination of mutation_rate, nb_elite and alphas possible nb_test_desired times. For each sets of parameters, results will be saved in a file named according to the set of parameter. One can regroup the results of each run in a csv file by using the print_csv function specifying the names of each file that needs to be treated and the number of performed runs on each parameter set
Usage
hyperparam_test_genetic_algorithm(
epochs,
nb_individuals,
ATCtree,
observations,
nb_test_desired,
mutation_rate,
nb_elite,
alphas,
path = "./",
num_thread = 1L
)
Arguments
epochs |
: the number of epochs for the genetic algorithm |
nb_individuals |
: the size of the population in the genetic algorithm |
ATCtree |
: ATC tree with upper bound of the DFS (without the root) |
observations |
: observation of the AE based on the medications of each patients (a DataFrame containing the medication on the first column and the ADR (boolean) on the second) on which we want to compute the risk distribution |
nb_test_desired |
: number of genetic algorithm runs on each sets of parameters |
mutation_rate |
: a vector with each mutation_rate to be tested |
nb_elite |
: a vector with each nb_elite to be tested |
alphas |
: a vector with each alphas to be tested |
path |
: the path where the resulting files should be written |
num_thread |
: Number of thread to run in parallel if openMP is available, 1 by default |
Value
No return value, this function should output results of the runs of the genetic algorithm in a specific format supported by function print_csv and p_value_csv_file. The files are outputed in path which is current directory by default.
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
# different parameter to test for
mutation_rate = c(.1,.2,.3)
nb_elite = c(0,1,2)
alphas = c(0.5,1,2)
hyperparam_test_genetic_algorithm(epochs = 2, nb_individuals = 2,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy,
nb_test_desired = 5, mutation_rate = mutation_rate,
nb_elite = nb_elite, alphas = alphas)
Function used to convert integer cocktails (like the one outputed by the distributionApproximation function) to string cocktail in order to make them more readable
Description
Function used to convert integer cocktails (like the one outputed by the distributionApproximation function) to string cocktail in order to make them more readable
Usage
int_cocktail_to_string_cocktail(cocktails, ATC_name)
Arguments
cocktails |
cocktails vector to be converted (index in the ATC tree) |
ATC_name |
The ATC_name column of the ATC tree |
Value
The name of integer cocktails in cocktails
Examples
data("ATC_Tree_UpperBound_2024")
int_list = list(c(561, 904),
c(1902, 4585))
int_cocktail_to_string_cocktail(int_list, ATC_Tree_UpperBound_2024$Name)
Used to add the p_value to each cocktail of cocktail list
Description
Used to add the p_value to each cocktail of cocktail list
Usage
p_value_cocktails(
distribution_outputs,
cocktails,
ATCtree,
observations,
num_thread = 1L,
filtred_distribution = FALSE
)
Arguments
distribution_outputs |
A list of distribution of cocktails of different sizes in order to compute the p_value for multiple cocktail sizes |
cocktails |
A list containing cocktails in the form of vector of integers (ATC index) |
ATCtree |
ATC tree with upper bound of the DFS (without the root) |
observations |
observation of the AE based on the medications of each patients (a DataFrame containing the medication on the first column and the ADR (boolean) on the second) on which we want to compute the risk distribution |
num_thread |
Number of thread to run in parallel if openMP is available, 1 by default |
filtred_distribution |
Does the p-values have to be computed using filtered distribution or normal distribution (filtered distribution by default) |
Value
A real valued number vector representing the p-value of the inputed cocktails computed on the distribution_outputs List.
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
DistributionApproximationResults_size2 = DistributionApproximation(epochs = 10,
ATCtree = ATC_Tree_UpperBound_2024, observations = FAERS_myopathy, Smax = 2)
DistributionApproximationResults_size3 = DistributionApproximation(epochs = 10,
ATCtree = ATC_Tree_UpperBound_2024, observations = FAERS_myopathy, Smax = 3)
score_distribution_list = list(DistributionApproximationResults_size2,
DistributionApproximationResults_size3)
cocktails = list(c(561, 904),
c(1902, 4585))
p_value_cocktails(score_distribution_list, cocktails, ATC_Tree_UpperBound_2024,
FAERS_myopathy)
Used to add the p_value to each cocktail of a csv_file that is an output of the genetic algorithm
Description
Used to add the p_value to each cocktail of a csv_file that is an output of the genetic algorithm
Usage
p_value_csv_file(
distribution_outputs,
filename,
filtred_distribution = FALSE,
sep = ";"
)
Arguments
distribution_outputs |
A list of distribution of cocktails of different sizes in order to compute the p_value for multiple cocktail sizes |
filename |
The file name of the .csv file containing the output |
filtred_distribution |
Does the p-values have to be computed using filtered distribution or normal distribution (filtered distribution by default) |
sep |
The separator used in the csv file (';' by default) |
Value
A real valued number vector representing the p-value of the inputed csv file filename, computed on the distribution_outputs List.
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
DistributionApproximationResults_size2 = DistributionApproximation(epochs = 10,
ATCtree = ATC_Tree_UpperBound_2024, observations = FAERS_myopathy, Smax = 2)
DistributionApproximationResults_size3 = DistributionApproximation(epochs = 10,
ATCtree = ATC_Tree_UpperBound_2024, observations = FAERS_myopathy, Smax = 3)
score_distribution_list = list(DistributionApproximationResults_size2,
DistributionApproximationResults_size3)
p_value_csv_file(score_distribution_list, "path/to/output.csv")
Used to add the p_value to each cocktail of an output of the genetic algorithm
Description
Used to add the p_value to each cocktail of an output of the genetic algorithm
Usage
p_value_genetic_results(
distribution_outputs,
genetic_results,
filtred_distribution = FALSE
)
Arguments
distribution_outputs |
A list of distribution of cocktails of different sizes in order to compute the p_value for multiple cocktail sizes |
genetic_results |
outputs of the genetic algorithm |
filtred_distribution |
Does the p-values have to be computed using filtered distribution or normal distribution (filtered distribution by default) |
Value
A real valued number vector representing the p-value of the inputed genetic algorithm results (genetic_results) computed on the distribution_outputs List.
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
DistributionApproximationResults_size2 = DistributionApproximation(epochs = 10,
ATCtree = ATC_Tree_UpperBound_2024, observations = FAERS_myopathy, Smax = 2)
DistributionApproximationResults_size3 = DistributionApproximation(epochs = 10,
ATCtree = ATC_Tree_UpperBound_2024, observations = FAERS_myopathy, Smax = 3)
score_distribution_list = list(DistributionApproximationResults_size2,
DistributionApproximationResults_size3)
genetic_results = GeneticAlgorithm(epochs = 10, nbIndividuals = 20,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy)
p_value_genetic_results(score_distribution_list, genetic_results)
Calculate p-value of sampled value
Description
Calculate p-value of sampled value
Usage
p_value_on_sampled(
empirical_distribution,
sampled_values,
isFiltered = FALSE,
includeZeroValue = FALSE
)
Arguments
empirical_distribution |
A numeric vector of values representing the empirical distribution (return value of DistributionAproximation function) |
sampled_values |
A scalar or a vector of real valued number representing the sampled value (score to be tested) |
isFiltered |
A boolean representing if we want to use the filtered distribution or the distribution as is (False by default) |
includeZeroValue |
A boolean that indicate if you want to take into account the null score (False by default) |
Value
A numeric value representing the empirical p-value
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
cocktails = list(c(561, 904),
c(1902, 4585))
estimated_score_distribution = DistributionApproximation(epochs = 10,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy)
Hypergeom_of_cocktails = compute_hypergeom_on_list(cocktails = cocktails,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy)
p_value = p_value_on_sampled(empirical_distribution = estimated_score_distribution,
sampled_values = Hypergeom_of_cocktails)
Plot the evolution of the mean and the best value of the population used by the GeneticAlgorithm
Description
Plot the evolution of the mean and the best value of the population used by the GeneticAlgorithm
Usage
plot_evolution(
list,
mean_color = "#F2A900",
best_color = "#008080",
xlab = "Epochs",
ylab = "Score"
)
Arguments
list |
A list with 2 elements returned by the GeneticAlgorithm: "mean" and "best", containing the numeric vectors representing the mean and best fitness of the population |
mean_color |
A string specifying the color of the mean values |
best_color |
A string specifying the color of the best values |
xlab |
A string specifying the label for the x-axis |
ylab |
A string specifying the label for the y-axis |
Value
no returned value, should plot the evolution of the genetic algorithm results (mean/max score for each epoch).
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
results = GeneticAlgorithm(epochs = 10, nbIndividuals = 10,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy)
plot_evolution(list = results)
Plot the histogram of the approximation of the RR distribution
Description
Plot the histogram of the approximation of the RR distribution
Usage
plot_frequency(
estimated,
sqrt = FALSE,
binwidth = 0.1,
hist_color = "#69b3a2",
density_color = "#FF5733",
xlab = "Score"
)
Arguments
estimated |
The ScoreDistribution element in the list returned by the DistributionApproximation function |
sqrt |
A Boolean to specify whether we normalize the estimated or not, it is recommended on large random walk. |
binwidth |
The width of the histogram bins |
hist_color |
The fill color for the histogram bars |
density_color |
The color for the density curve |
xlab |
Label of X axis |
Value
no returned value, should plot the histogram of the estimated distribution (estimated).
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
estimation = DistributionApproximation(epochs = 10, ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy)
plot_frequency(estimated = estimation$ScoreDistribution)
Print every cocktails found during the genetic algorithm when used with the hyperparam_test_genetic_algorithm function. This enables to condense the solutions found in each files by collapsing similar cocktail in a single row by cocktail.
Description
Print every cocktails found during the genetic algorithm when used with the hyperparam_test_genetic_algorithm function. This enables to condense the solutions found in each files by collapsing similar cocktail in a single row by cocktail.
Usage
print_csv(
input_filenames,
observations,
repetition,
ATCtree,
csv_filename = "solutions.csv"
)
Arguments
input_filenames |
: A List containing filename of hyperparam_test_genetic_algorithm output file |
observations |
: observation of the AE based on the medications of each patients (a DataFrame containing the medication on the first column and the ADR (boolean) on the second) on which we want to compute the risk distribution |
repetition |
: The parameter nb_test_desired used in the hyperparam test function |
ATCtree |
: ATC tree with upper bound of the DFS (without the root) |
csv_filename |
: Name of the output file, "solutions.csv" by default |
Value
No return value, should process the output of the genetic algorithm in files produced by hyperparam_test_genetic_algorithm and output a summary csv file. The csv file is outputed in current directory and named after the csv_filename variable (solutions.csv by default).
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
files = c('250e_700ind_0.2mr_0ne_2alpha.txt') # results of hyperparam_test_genetic_algorithm
print_csv(input_filenames = files, observations = FAERS_myopathy,
repetition = 5, ATCtree = ATC_Tree_UpperBound_2024)
Make a Quantile-Quantile diagram from the output of the MCMC algorithm (DistributionAproximation) and the algorithm that exhaustively calculates the distribution
Description
Make a Quantile-Quantile diagram from the output of the MCMC algorithm (DistributionAproximation) and the algorithm that exhaustively calculates the distribution
Usage
qq_plot_output(estimated, true, filtered = FALSE, color = "steelblue")
Arguments
estimated |
Outputed object of DistributionApproximation function |
true |
Outputed object of either DistributionApproximation function or True distribution computation function |
filtered |
Make use of the classic distributuion estimation or of the filtred one (number of patient taking the cocktail > beta) |
color |
The color of the dashed line of the qq-plot |
Value
no returned value, should plot the quantile-quantile plot of the estimated distribution (estimated) vs the true distribution (true).
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
estimated_score_distribution = DistributionApproximation(epochs = 10,
ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy[1:100,], Smax =2)
true_score_distribution = trueDistributionSizeTwoCocktail(ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy[1:100,], beta = 4)
qq_plot_output(estimated = estimated_score_distribution,
true = true_score_distribution)
Function used to convert a string vector of drugs in form "drug1:drug2" to a vector of index of the ATC tree ex: c(ATC_index(drug1), ATC_index(drugs2))
Description
Function used to convert a string vector of drugs in form "drug1:drug2" to a vector of index of the ATC tree ex: c(ATC_index(drug1), ATC_index(drugs2))
Usage
string_list_to_int_cocktails(ATC_name, lines)
Arguments
ATC_name |
the ATC_name column of the ATC tree |
lines |
A string vector of drugs cocktail in the form "drug1:drug2:...:drug_n" |
Value
An R List that can be used by other algorithms (e.g. clustering algorithm)
Examples
data("ATC_Tree_UpperBound_2024")
string_list = c('hmg coa reductase inhibitors:nervous system',
'metformin:prasugrel')
string_list_to_int_cocktails(ATC_Tree_UpperBound_2024$Name,
string_list)
The true distribution of the score among every single nodes of the ATC
Description
The true distribution of the score among every single nodes of the ATC
Usage
trueDistributionDrugs(
ATCtree,
observations,
beta,
max_score = 1000L,
nbResults = 100L,
num_thread = 1L
)
Arguments
ATCtree |
: ATC tree with upper bound of the DFS (without the root) |
observations |
: observation of the AE based on the medications of each patients (a DataFrame containing the medication on the first column and the ADR (boolean) on the second) on which we want to compute the risk distribution |
beta |
: minimum number of person taking the cocktails in order to consider it in the beta score distribution |
max_score |
: maximum number the score can take. Score greater than this one would be added to the distribution as the value max_score. Default is 1000 |
nbResults |
: Number of returned solution (Cocktail with the best oberved score during the run), 100 by default |
num_thread |
: Number of thread to run in parallel if openMP is available, 1 by default |
Value
Return a List containing : - ScoreDistribution : the distribution of the score as an array with each cells representing the number of risks = (index-1)/ 10 - Filtered_score_distribution : Distribution containing score for cocktails taken by at least beta patients. - Outstanding_score : An array of the score greater than max_score, - Best_cocktails : the nbResults bests cocktails encountered during the run. - Best_cocktails_beta : the nbResults bests cocktails taken by at least beta patients encountered during the run. - Best_scores : Score corresponding to the Best_cocktails. - Best_scores_beta : Score corresponding to the Best_cocktails_beta.
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
size_1_score_distribution = trueDistributionDrugs(ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy[1:100,], beta = 4)
The true distribution of the score among every size-two cocktails
Description
The true distribution of the score among every size-two cocktails
Usage
trueDistributionSizeTwoCocktail(
ATCtree,
observations,
beta,
max_score = 100L,
nbResults = 100L,
num_thread = 1L
)
Arguments
ATCtree |
: ATC tree with upper bound of the DFS (without the root) |
observations |
: observation of the AE based on the medications of each patients (a DataFrame containing the medication on the first column and the ADR (boolean) on the second) on which we want to compute the risk distribution |
beta |
: minimum number of person taking the cocktails in order to consider it in the beta score distribution |
max_score |
: maximum number the score can take. Score greater than this one would be added to the distribution as the value max_score. Default is 1000 |
nbResults |
: Number of returned solution (Cocktail with the best oberved score during the run), 100 by default |
num_thread |
: Number of thread to run in parallel if openMP is available, 1 by default |
Value
Return a List containing : - ScoreDistribution : the distribution of the score as an array with each cells representing the number of risks = (index-1)/ 10 - Filtered_score_distribution : Distribution containing score for cocktails taken by at least beta patients. - Outstanding_score : An array of the score greater than max_score, - Best_cocktails : the nbResults bests cocktails encountered during the run. - Best_cocktails_beta : the nbResults bests cocktails taken by at least beta patients encountered during the run. - Best_scores : Score corresponding to the Best_cocktails. - Best_scores_beta : Score corresponding to the Best_cocktails_beta.
Examples
data("ATC_Tree_UpperBound_2024")
data("FAERS_myopathy")
size_2_score_distribution = trueDistributionSizeTwoCocktail(ATCtree = ATC_Tree_UpperBound_2024,
observations = FAERS_myopathy[1:100,], beta = 4)