generate_test_stat_hist: Generate the test statistic and p-values under the null...

In GeneAccord: Detection of clonally exclusive gene or pathway pairs in a cohort of cancer patients

Description

Generate the values of the test statistic under the null, and also p-values of the clonal exclusivity test under the null. Taking the average rates of clonal exclusivity, as well as sampling from the real data for each patient, in how many trees a pair occurs and is clonally exclusive.

Usage

generate_test_stat_hist(avg_rates_m, list_of_num_trees_all_pats,
  list_of_clon_excl_all_pats, ecdf_list, num_pat_pair_max, num_pairs_sim,
  beta_distortion = 1000)

Arguments

avg_rates_m	The average rates of clonal exclusivity from all the patients in the cohort, and averaged over several trees from the collection of tree inferences.
list_of_num_trees_all_pats	A named list that contains an entry for each patient which is the vector with the values of the information from each pair in a patient of how often it was mutated across trees. The patient odering in the list has to be the same as in avg_rates_m.
list_of_clon_excl_all_pats	A named list with an entry for each patient that is a vector with the values of in how many trees a pair was clonally exclusive. The patient ordering in the list has to be the same as in avg_rates_m.
ecdf_list	The list with ECDF's as generated with generate_ecdf_test_stat.
num_pat_pair_max	The maximum number of patients a pair is mutated in.
num_pairs_sim	The number of simulated gene/pathway pairs to be generated, i.e. the number of times the test statistic is computed.
beta_distortion	The value M=alpha + beta for the beta distribution, with which the average rates will be distorted. The bigger the M the higher the distribution is peaked around the actual rate. Therefore, the lesser the M, the more distorted the rates will be. Default: 1000.

Details

This function takes the computed average rates of clonal exclusivity from the data (m1, ... mN), which are specific to each patient and averaged over several trees from the collection of tree inferences. It also takes the histogram for each patient, of the values of how often a pair was clonally exclusive over the number of trees it was mutated in. It also takes the empirical cumulative distribution function (ECDF) which was generated with generate_ecdf_test_stat. It then computes the p-value of the simulated pairs under the null.

Value

The return value is a list of tibbles with a tibble for each number of patients, a pair can be mutated in. Each tibble contains the columns 'test_statistic', 'mle_delta', and then num_pat_pair columns of the rates of each patient 'pat1', 'pat2', ...; as well as num_pat_pair columns with the information about each patient, in how many trees the pair was occurring and in how many trees the pair was clonally exclusive. The tibble also contains a column 'pval' with the p-value of the simulated pair. The list of tibbles is of length minnum_pat_pair_max, length(avg_rates_m).

Author(s)

Ariane L. Moore

Examples

clone_tbl <- dplyr::tibble("file_name" =
       rep(c(rep(c("fn1", "fn2"), each=3)), 2),
       "patient_id"=rep(c(rep(c("pat1", "pat2"), each=3)), 2),
       "altered_entity"=c(rep(c("geneA", "geneB", "geneC"), 4)),
       "clone1"=c(0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 0, 0),
       "clone2"=c(1, 0, 1, 0, 1, 1, 1, 0, 0, 1, 0, 1),
       "tree_id"=c(rep(5, 6), rep(10, 6)))
clone_tbl_pat1 <- dplyr::filter(clone_tbl, patient_id == "pat1")
clone_tbl_pat2 <- dplyr::filter(clone_tbl, patient_id == "pat2")
rates_exmpl_1 <- compute_rates_clon_excl(clone_tbl_pat1)
rates_exmpl_2 <- compute_rates_clon_excl(clone_tbl_pat2)
avg_rates_m <- apply(cbind(rates_exmpl_1, rates_exmpl_2), 2, mean)
names(avg_rates_m) <- c(names(rates_exmpl_1)[1], names(rates_exmpl_2)[1])
values_clon_excl_num_trees_pat1 <- get_hist_clon_excl(clone_tbl_pat1)
values_clon_excl_num_trees_pat2 <- get_hist_clon_excl(clone_tbl_pat2)
list_of_num_trees_all_pats <-
 list(pat1=values_clon_excl_num_trees_pat1[[1]], 
      pat2=values_clon_excl_num_trees_pat2[[1]])
list_of_clon_excl_all_pats <- 
  list(pat1=values_clon_excl_num_trees_pat1[[2]],
       pat2=values_clon_excl_num_trees_pat2[[2]])
num_pat_pair_max <- 2
num_pairs_sim <- 10
ecdf_list <- generate_ecdf_test_stat(avg_rates_m, 
                           list_of_num_trees_all_pats, 
                           list_of_clon_excl_all_pats,
                           num_pat_pair_max, num_pairs_sim)
sim_res <- generate_test_stat_hist(avg_rates_m, 
                                  list_of_num_trees_all_pats, 
                                  list_of_clon_excl_all_pats, 
                                  ecdf_list, 
                                  num_pat_pair_max, 
                                  num_pairs_sim)

GeneAccord documentation built on Nov. 8, 2020, 8:04 p.m.