|
| 1 | +prepare_data <- function(input_file){ |
| 2 | + suppressMessages(require(dplyr)) |
| 3 | + suppressMessages(require(stringr)) |
| 4 | + suppressMessages(require(AnnotationDbi)) |
| 5 | + suppressMessages(require(org.Hs.eg.db)) |
| 6 | + |
| 7 | + dge <- read.table(file = input_file, |
| 8 | + sep = "\t", |
| 9 | + header = T, |
| 10 | + check.names = F) |
| 11 | + # row.names = 1) |
| 12 | + |
| 13 | + #--------id analysis-------- |
| 14 | + |
| 15 | + if ("translated_names" %in% colnames(dge)) { |
| 16 | + df <- dge %>% dplyr::select(c(translated_names, log2FoldChange, padj)) |
| 17 | + df$geneid = mapIds(org.Hs.eg.db, |
| 18 | + keys=as.character(df$translated_names), |
| 19 | + column="ENTREZID", |
| 20 | + keytype="SYMBOL", |
| 21 | + multiVals="first") |
| 22 | + df2 <- df %>% dplyr::filter(!str_detect(translated_names, "ENSG0")) %>% dplyr::filter(!str_detect(geneid, "NA")) |
| 23 | + } else { |
| 24 | + df <- dge %>% dplyr::select(c(gene_name, log2FoldChange, padj)) |
| 25 | + df <- df %>% dplyr::filter(padj < 0.05) |
| 26 | + df$geneid = mapIds(org.Hs.eg.db, |
| 27 | + keys=as.character(df$gene_name), |
| 28 | + column="ENTREZID", |
| 29 | + keytype="SYMBOL", |
| 30 | + multiVals="first") |
| 31 | + df2 <- df %>% dplyr::filter(!str_detect(gene_name, "ENSG0")) %>% dplyr::filter(!str_detect(geneid, "NA")) |
| 32 | + } |
| 33 | + |
| 34 | + input_data <- as.data.frame(df2) |
| 35 | + return(input_data) |
| 36 | +} |
| 37 | + |
| 38 | +# gage_pathview <- function(input, kegg_pathway, pgeomean, output_folder){ |
| 39 | +gage_pathview <- function(input_file, pgeomean = 0.25, output_folder = output_folder){ |
| 40 | + |
| 41 | + message("Gage/Pathview Analysis Start") |
| 42 | + source("KEGG_gene_annotation.R") |
| 43 | + |
| 44 | + #--------Libraries-------- |
| 45 | + message("Loading Libraries") |
| 46 | + suppressMessages(require(pathview)) |
| 47 | + suppressMessages(require(gage)) |
| 48 | + |
| 49 | + #--------Output directory-------- |
| 50 | + message("Creating Output Directory") |
| 51 | + |
| 52 | + mainDir <- setwd(getwd()) |
| 53 | + outDir <- file.path(mainDir) |
| 54 | + subDir <- output_folder |
| 55 | + resDir <- file.path(outDir, subDir) |
| 56 | + |
| 57 | + if (file.exists(resDir)){ |
| 58 | + message ("Results folder for analysis found") |
| 59 | + } else { |
| 60 | + dir.create(resDir) |
| 61 | + message ("Results folder for analysis created") |
| 62 | + } |
| 63 | + |
| 64 | + #--------load data-------- |
| 65 | + message("Loading data and annotating Ids") |
| 66 | + |
| 67 | + input_data <- prepare_data(input_file) |
| 68 | + # return(input_data) |
| 69 | + |
| 70 | + #--------load Kegg pathways-------- |
| 71 | + message("Loading Kegg Pathways") |
| 72 | + |
| 73 | + ## pathways from GAGE kegg.gsets |
| 74 | + pathways <- kegg.gsets(species = "hsa", |
| 75 | + id.type = "kegg", |
| 76 | + check.new = T) |
| 77 | + # sigmetidx <- pathways$sigmet.idx |
| 78 | + # listkegg <- pathways$kg.sets[sigmetidx] |
| 79 | + listkegg <- pathways$kg.sets |
| 80 | + |
| 81 | + ## prepare vector with geneid + FC |
| 82 | + foldchanges = input_data$log2FoldChange |
| 83 | + names(foldchanges) = input_data$geneid |
| 84 | + # head(foldchanges) |
| 85 | + |
| 86 | + #--------run Gage------- |
| 87 | + message("Gage Analysis") |
| 88 | + |
| 89 | + keggres = gage(exprs = foldchanges, |
| 90 | + gsets = listkegg, |
| 91 | + same.dir = T) |
| 92 | + |
| 93 | + lapply(keggres, head) |
| 94 | + |
| 95 | + keggresp <- as.data.frame(keggres$greater) |
| 96 | + keggresp$pathway <- rownames(keggres$greater) |
| 97 | + keggresp2 <- keggresp %>% dplyr::filter(p.geomean < pgeomean) |
| 98 | + |
| 99 | + keggrespathways = data.frame(id=keggresp2$pathway, keggresp2) %>% |
| 100 | + tbl_df() %>% |
| 101 | + # filter(row_number()<=5) %>% |
| 102 | + .$id %>% |
| 103 | + as.character() |
| 104 | + |
| 105 | + keggresids = substr(keggrespathways, start=1, stop=8) |
| 106 | + # avoid hsa01100 kegg pathway (too big) |
| 107 | + keggresids <- keggresids[!keggresids %in% "hsa01100"] |
| 108 | + |
| 109 | + #--------Pathview analysis-------- |
| 110 | + message("Pathview Analysis") |
| 111 | + |
| 112 | + # detach dplyr (in order to avoid issues) as reported here: https://www.biostars.org/p/242157/ |
| 113 | + detach("package:dplyr", unload=TRUE) |
| 114 | + |
| 115 | + # set output |
| 116 | + setwd(output_folder) |
| 117 | + |
| 118 | + # Define plotting function for applying later |
| 119 | + plot_pathway = function(pid) pathview(gene.data=foldchanges, pathway.id=pid, species="hsa", new.signature=FALSE) |
| 120 | + |
| 121 | + # plot multiple pathways (plots saved to disk and returns a throwaway list object) |
| 122 | + tmp = sapply(keggresids, function(pid) pathview(gene.data=foldchanges, pathway.id=pid, species="hsa")) |
| 123 | + |
| 124 | + #---------Saving results---------------- |
| 125 | + message("Saving Results on Disk and as Rds object") |
| 126 | + |
| 127 | + write.table(keggresp2, "GageResults.tab", sep = "\t", row.names = F) |
| 128 | + |
| 129 | + report_data <- list() |
| 130 | + |
| 131 | + # save results (for Rmd report) |
| 132 | + report_data[["DataFrame"]] <- as.data.frame(input_data) |
| 133 | + report_data[["FoldChanges"]] <- foldchanges |
| 134 | + report_data[["Pgeomean"]] <- pgeomean |
| 135 | + report_data[["PathviewResults"]] <- tmp |
| 136 | + report_data[["GageResults"]] <- keggresp2 |
| 137 | + |
| 138 | + # saveRds (tmp) |
| 139 | + saveRDS(report_data, "GagePathviewAnalysis.Rds") |
| 140 | + message("Gage/Pathview analysis completed") |
| 141 | + |
| 142 | +} |
0 commit comments