add hypergeo tests

R/0docs.R

@@ -19,6 +19,8 @@
 #' @param width Width of the saved plot.
 #' @param heights Passed to \link[patchwork]{wrap_plots}.
 #' @param subtitle_size Size of the plot subtitle.
+#' @param phenotype_to_genes Phenotype to gene mapping from
+#' \link[HPOExplorer]{load_phenotype_to_genes}.
 #' @family plot_
 #' @returns R object.
 #' @name plot_

R/plot_ontology_levels.R

@@ -13,6 +13,7 @@
 #' @param log_vars Logical vector indicating which variables to log-transform.
 #' @param sig_vars Logical vector indicating which variables to only plot
 #' for significant results.
+#' @param return_data Return the full long data used in the plots.
 #' @inheritParams plot_
 #' @inheritParams ggpubr::stat_cor
 #' @inheritParams prioritise_targets
@@ -57,7 +58,8 @@ plot_ontology_levels <- function(results = load_example_results(),
                          height=7,
                          width=length(x_vars)*5.75,
                          smooth.line.args=list(method = "loess",
-                                               se = FALSE)
+                                               se = FALSE),
+                         return_data=TRUE
                          ){
 
   requireNamespace("ggplot2")
@@ -252,6 +254,7 @@ plot_ontology_levels <- function(results = load_example_results(),
                           height = height,
                           width = width)
   }
+  if(isFALSE(return_data)) r2 <- NULL
   return(list(data=r2,
               data_stats=data_stats,
               plot=plts2))

R/plot_report.R

@@ -5,8 +5,6 @@
 #' @param rep_dt Report table.
 #' @param annot HPO annotations.
 #' @param remove_cols Columns to remove from \code{rep_dt}.
-#' @param phenotype_to_genes Phenotype to gene mapping from
-#' \link[HPOExplorer]{load_phenotype_to_genes}.
 #' @inheritParams plot_
 #' @inheritParams ggnetwork_plot_full
 #' @inheritDotParams ggplot2::ggsave

R/prioritise_targets.R

@@ -209,6 +209,10 @@ prioritise_targets <- function(#### Input data ####
     Severity_score <- cl_name <- cl_id <- Severity_score_max <-
     info_content <- NULL;
 
+  force(results)
+  force(ctd_list)
+  force(hpo)
+  force(phenotype_to_genes)
   t1 <- Sys.time()
   messager("Prioritising gene targets.",v=verbose)
   #### Add logFC ####

R/ttd_check.R

@@ -34,18 +34,9 @@ ttd_check <- function(top_targets,
                       show_plot = TRUE,
                       save_path = NULL,
                       height=NULL,
-                      width=NULL){
-  # top_targets <- prioritise_targets(
-  #   keep_deaths = NULL,
-  #   keep_tiers = NULL,
-  #   severity_threshold_max = NULL,
-  #   severity_threshold = NULL,
-  #   pheno_frequency_threshold = NULL,
-  #   keep_onsets = NULL,
-  #   keep_ont_levels = seq(4),
-  #   keep_celltypes = NULL,
-  #   top_n = 2,
-  #   group_vars = c("hpo_id","disease_id"))$top_targets
+                      width=NULL,
+                      phenotype_to_genes=HPOExplorer::load_phenotype_to_genes()){
+  # top_targets <- prioritise_targets()$top_targets
   # drug_types <- c("Gene therapy"
     # "Antisense drug",
     # "Antisense oligonucleotide",
@@ -109,6 +100,14 @@ ttd_check <- function(top_targets,
   # length(unique(paste0(dat_sub2$DRUGID,dat_sub2$INDICATI,
   #                      dat_sub2$GENENAME2)))
   dat_sub[,prioritised:=(DRUGID %in% dat_sub2$DRUGID)]
+  #### Hypergeometric test ####
+  dat_sub[,failed:=HIGHEST_STATUS %in% failed_status]
+  fail <- dat_sub[failed==TRUE,drop=FALSE]
+  notfail <- dat_sub[failed==FALSE,drop=FALSE]
+  ttd_hypergeo_out <- ttd_hypergeo(fail=fail,
+                                   notfail=notfail,
+                                   top_targets=top_targets,
+                                   p2g=phenotype_to_genes)
   #### Plot ####
   plt <- plot_ttd(dat_sub = dat_sub,
                   failed_status = failed_status)
@@ -124,6 +123,7 @@ ttd_check <- function(top_targets,
     list(data=dat_sub,
          data_overlap=dat_sub2,
          pct_captured,
-         plot=plt)
+         plot=plt,
+         ttd_hypergeo_out=ttd_hypergeo_out)
   )
 }

R/ttd_hypergeo.R

@@ -0,0 +1,37 @@
+ttd_hypergeo <- function(fail,
+                         notfail,
+                         top_targets,
+                         p2g=HPOExplorer::load_phenotype_to_genes()){
+  # https://seqqc.wordpress.com/2019/07/25/how-to-use-phyper-in-r/
+
+  ## Test for over-representation (enrichment)
+  # phyper(Overlap-1, group2, Total-group2, group1,lower.tail= FALSE)
+  overlap <- data.table::uniqueN(notfail[prioritised==TRUE]$GENENAME3)
+  group2 <- data.table::uniqueN(notfail$GENENAME3)
+  total <- data.table::uniqueN(c(p2g$gene_symbol,notfail$GENENAME3))
+  group1 <- data.table::uniqueN(top_targets$gene_symbol)
+  nonfailed_enrichment <- stats::phyper(overlap-1,
+                                        group2,
+                                        total-group2,
+                                        group1,
+                                        lower.tail= FALSE)
+  messager("Non-failed gene targets enrichment p-value:",nonfailed_enrichment)
+  ## Test for under-representation (depletion)
+  # phyper(Overlap, group2, Total-group2, group1, lower.tail= TRUE)
+  overlap <- data.table::uniqueN(fail[prioritised==TRUE]$GENENAME3)
+  group2 <- data.table::uniqueN(fail$GENENAME3)
+  total <- data.table::uniqueN(c(p2g$gene_symbol,fail$GENENAME3))
+  group1 <- data.table::uniqueN(top_targets$gene_symbol)
+  failed_depletion <- stats::phyper(overlap,
+                                    group2,
+                                    total-group2,
+                                    group1,
+                                    lower.tail= TRUE)
+  messager("Failed gene targets depletion p-value:",failed_depletion)
+  return(
+    list(
+      nonfailed_enrichment=nonfailed_enrichment,
+      failed_depletion=failed_depletion
+    )
+  )
+}