GeneAccord: Detection of clonally exclusive gene or pathway pairs in a...

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

Description

Method to detect clonally exclusive gene or pathway pairs in a cohort of cancer patients

Usage

GeneAccord(clone_tbl, avg_rates_m, ecdf_list, alternative = "greater",
  genes_of_interest = "ALL", AND_OR = "OR")

Arguments

clone_tbl	The tibble containing the information of which gene/pathway is mutated in which clone from which patient and in which tree from the collection of trees. Can be generated with create_tbl_tree_collection for each patient separately and then appended.
avg_rates_m	The average rates of clonal exclusivity for each patient as computed with compute_rates_clon_excl. The name of each rate is the respective patient id. The rates are assumed to be the average over all tree inferences from a patient.
ecdf_list	The list of ECDF's of the test statistic under the null distribution. Can be generated with generate_ecdf_test_stat.
alternative	The character indicating whether pairs should only be tested if delta > 0 or if all pairs should be tested. Can be one of "greater" or "two.sided". Default: "greater".
genes_of_interest	A character vector of genes to test for clonal exclusivity. The genes have to be in the same identifier as the one in the tibble. Per default, all genes are tested. Default: "ALL".
AND_OR	If genes_of_interest is specified, this indicator tells whether to test only pairs within the genes_of_interest (AND), or whether all pairs involving at least one of these genes should be tested (OR). I.e. can be one of "AND", "OR". Default: "OR". If genes_of_interest is "ALL", then all gene pairs will be tested and this parameter is ignored.

Details

After running a tool such as Cloe that identifies clones in a tumor and infers the phylogenetic history, the user has for each tumor a list of alterations and their clone assignments. Since the tree inference includes uncertainty, it may be run several times. Given a tibble containing the information of which genes/pathways are mutated in which patient and clone and from which tree, this function systematically tests the data for significant clonal exclusivities. That is, it checks for each gene/pathway pair whether the number of clonal exclusivities is significantly different from what would be expected by chance. Such a tibble can be generated with create_tbl_tree_collection, and then adding the additional column 'tree_id' to indicate which tree of the tree inference was used. For instance, if the tree inference tool was run several times using different seeds, the column 'tree_id' may contain the seed of the respective tree. Hence, the tibble is expected to have the columns 'file_name', 'patient_id', 'altered_entity', 'clone1', 'clone2', ... up to the maximal number of clones (Default: until 'clone7'), and 'tree_id'. Note that the labelling of the clones does not matter and only needs to stay fixed within each patient and tree inference. There is also the option to test two-sided, meaning that also pairs will be tested that tend to occur more often together in the same clones or separate in different clones. Hence it also allows to detect significant clonal co-occurrence. An additional option is to test only a specific subset of genes.

Value

A tibble containing the test result for each pair of mutated genes/pathways that was tested. More precisely, it contains the columns 'entity_A', 'entity_B', 'num_patients', 'pval', 'mle_delta', 'test_statistic', and 'qval'. Each row is then a gene or pathway pair which is specified with 'entity_A', and 'entity_B'. Note that the test is symmetric, hence switching the labels A and B does not change the results. The column 'num_patients' contains the information in how many patients both of the genes/pathways were mutated and hence how many patients' rates were used for the test. The 'pval' is the p-value of the clonal exclusivity test. The 'mlde_delta' is the maximum likelihood estimate of the delta for the elevated clonal exclusivity rate in the alternative model. The column 'test_statistic' is the likelihood ratio test statistic. The 'qval' is the adjusted p-value after multiple testing correction with Benjamini-Hochberg.

Author(s)

Ariane L. Moore, [e-mail]

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)
alternative <- "greater"
GeneAccord(clone_tbl, avg_rates_m, ecdf_list, alternative)
alternative <- "two.sided"
GeneAccord(clone_tbl, avg_rates_m, ecdf_list, alternative)
genes_of_interest <- c("geneB", "geneC")
GeneAccord(clone_tbl, avg_rates_m, ecdf_list, 
            alternative, genes_of_interest)
AND_OR <- "AND"
GeneAccord(clone_tbl, avg_rates_m, ecdf_list, 
            alternative, genes_of_interest, AND_OR)

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