enrich {AEenrich}

Perform Adverse Event Enrichment Tests

Description

The enrich function is used to perform Adverse event (AE) enrichment analysis. Unlike the continuous gene expression data, AE data are counts. Therefore, AE data has many zeros and ties. We propose two enrichment tests. AEFisher is a modified Fisher's exact test based on pre-selected significant AEs, while AEKS is based on a modified Kolmogorov-Smirnov statistic.

Usage

enrich(
  data,
  dd.group,
  drug.case,
  drug.control = NULL,
  method = "aeks",
  n_perms = 1000,
  covar = NULL,
  p = 0,
  q.cut = 0.1,
  or.cut = 1.5,
  zero = FALSE,
  min_size = 5,
  min_AE = 10,
  cores = detectCores()
)

Arguments

data

a data.frame. Two data types are allowed. Type I data consisting data on the report level, having ID, Drug type and AE name as the first 3 columns with covariates(optional) followed. Type II data have drug type and AE name as the first two columns, with the 3rd and 4th Columns giving the numbers of successes(have AE) and failures(Do not have AE) respectively, then followed by covariates. See example data for details.

dd.group

a data.frame with AE name and Group name. This data.frame have the group information for each individual AE.

drug.case

a character string for the target drug of interest.

drug.control

a character string for the reference drug. If NULL(default), all other drugs combined are the reference.

method

a character string specifying the method for the enrichment test. It must take "aeks" (default) or "aefisher"; "aeks" is the rank-based enrichment test, and "aefisher" is the Fisher enrichment test. See details described in the paper (see reference section of this document).

n_perms

an integer value specifying the number of permutations in permutation test.

covar

a character vector specifying the columns of covariates, default NULL.

p

a numerical value to control the weight of the step, can take any value between 0 and 1. If 0(default), reduces to the standard Kolmogorov-Smirnov statistics.

q.cut

a numerical value specifying the significance cut for q value of AEs in aefisher.

or.cut

a numerical value specifying the significance cut for odds ratio of AEs in aefisher.

zero

logical, default FALSE.If TRUE, add zero indicator to enrichment score.

min_size

the minimum size of group required for enrichment analysis.

min_AE

the minimum number of cases required to start counting for a specific AE.

cores

the number of cores to use for parallel execution.

Value

A list containing 2 data.frames named Final_result and AE_info.

The Final_result data.frame contains the following columns:

• GROUP_NAME: AE group names

• ES: enrichment score

• p_value: p value of the enrichment test

• GROUP_SIZE: number of AEs per group

The AE_info contains the following columns:

• AE_NAME: AE names

• OR: odds ratio for each individual AE

• p_value: p value for AE-drug association

• 95Lower: lower bound of 95 percent confidence interval of odds ratio

• 95Lower: upper bound of 95 percent confidence interval of odds ratio

• se(logOR): standard error of log odds ratio

References

Li, S. and Zhao, L. (2020). Adverse event enrichment tests using VAERS. arXiv:2007.02266.

Subramanian, A.e.a. (2005). Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. Proceedings of the National Academy of Sciences. 102. 15545-15550.

Tian, Lu & Greenberg, Steven & Kong, Sek Won & Altschuler, Josiah & Kohane, Isaac & Park, Peter. (2005). Discovering statistically significant pathways in expression profiling studies. Proceedings of the National Academy of Sciences of the United States of America. 102. 13544-9. 10.1073/pnas.0506577102.

Examples

# AEKS

### Type I data: data on report level
# enrich(data = covid1, covar = c("SEX", "AGE"), p = 0, method = "aeks",
#        n_perms = 1000, drug.case = "COVID19", dd.group = group, cores = 2,
#        drug.control = "OTHER", min_size = 5, min_AE = 10, zero = FALSE)
      
## Type II data: aggregated data
# enrich(data = covid2, covar = c("SEX", "AGE"), p = 0, method = "aeks",
#        n_perms = 1000, drug.case = "DrugYes", dd.group = group, cores = 2,
#        drug.control = "DrugNo", min_size = 5, min_AE = 10)
      
# AEFISHER
## Type I data: data on report level
# enrich(data = covid1, covar = c("SEX", "AGE"), p = 0, method = "aefisher",
#        n_perms = 1000, drug.case = "COVID19", dd.group = group,
#        drug.control = "OTHER", min_size = 5, min_AE = 10, q.cut = 0.05, 
#        or.cut = 1.5, cores = 2)
      
## Type II data: aggregated data
# enrich(data = covid2, covar = c("SEX", "AGE"), p = 0, method = "aefisher",
#        n_perms = 1000, drug.case = "DrugYes", dd.group = group,
#        drug.control = "DrugNo", min_size = 5, min_AE = 10, cores = 2)