signed enricher, gsea; gsea now accepts full transcriptome

Enricher previously (correctly) accepted the top 10 quantile and bottom
10 quantile of significant genes. It now does the same thing but with
the added constraint of genes actually having positive or negative logFC
with upregulated and downregulated pathways respectively.

GSEA previously (incorrectly) accepted the top 10 quantile and bottom 10
quantile of all genes; it should have been taking the entire
transcriptome. It now no longer accepts quantile measures for lower or
upper but instead calculates what is positive and what is negative in
terms of log fold change and then inputs that into GSEA.

Author: rela-v

R/do_enrichr_pod.R

@@ -43,10 +43,12 @@ do_enrichr_pod <- function(bpn, alpha = 0.05, lower = NULL, upper = NULL, useFDR
                "Cellular_Component")
 
   upreg <- inp %>%
+    filter(.data$Value_LogDiffExp >= 0) %>%
     filter(.data$Value_LogDiffExp >= upper) %>%
     dplyr::filter(.data$Significane_pvalue <= alpha)
 
   downreg <- inp %>%
+    filter(.data$Value_LogDiffExp <= 0) %>%
     filter(.data$Value_LogDiffExp <= lower) %>%
     dplyr::filter(.data$Significane_pvalue <= alpha)
 

R/do_gsea_pod.R

@@ -45,8 +45,9 @@ load_pathways <- function(species) {
 #' TRUE
 
 do_gsea_pod <- function(bpn, species = "hsapiens", gmtfile = NULL,
-                        lower = NULL, upper = NULL, alpha = 0.05,
-                        minSize = 15, maxSize = 500, rankWithSignificance = FALSE, useFDR=TRUE) {
+                        alpha = 0.05, minSize = 15, maxSize = 500,
+                        rankWithSignificance = FALSE, useFDR=TRUE) {
+
   if(!is.null(gmtfile)) {
     pathways <- fgsea::gmtPathways(gmtfile)
   } else if(!is.null(species)) {
@@ -57,19 +58,11 @@ do_gsea_pod <- function(bpn, species = "hsapiens", gmtfile = NULL,
 
   inp <- input(bpn)
 
-  if (is.null(lower)) {
-    lower <- round(stats::quantile(inp$Value_LogDiffExp, 0.10), 3)
-  }
-
-  if (is.null(upper)) {
-    upper <- round(stats::quantile(inp$Value_LogDiffExp, 0.90), 3)
-  }
-
   upreg <- inp %>%
-    dplyr::filter(.data$Value_LogDiffExp > upper)
+    dplyr::filter(.data$Value_LogDiffExp > 0)
 
   downreg <- inp %>%
-    dplyr::filter(.data$Value_LogDiffExp < lower)
+    dplyr::filter(.data$Value_LogDiffExp < 0)
 
   ranked <- inp %>%
     dplyr::mutate(score = (.data$Value_LogDiffExp * ifelse(rankWithSignificance, -log10(.data$Significane_pvalue), 1))) %>%