Add code for paper

Author: jaydu1

paper/AD/GO.R

@@ -0,0 +1,81 @@
+# install.packages("BiocManager")
+# BiocManager::install("clusterProfiler")
+# BiocManager::install("org.Hs.eg.db")
+# BiocManager::install("AnnotationDbi")
+# BiocManager::install("rrvgo")
+
+
+library(clusterProfiler)
+library(org.Hs.eg.db)
+library(AnnotationDbi)
+library(rrvgo)
+
+library(purrr)
+library(stringr)
+library(ggplot2)
+library(patchwork)
+library(dplyr)
+
+
+celltype <- 'exneu'
+args = commandArgs(trailingOnly=TRUE)
+cat(args)
+ps_model <- as.character(args[1])
+ps_model <- 'random_forest_cv'
+
+path_rs_GO <- sprintf('~/../jinandmaya/AD/results/')
+dir.create(path_rs_GO, recursive=TRUE, showWarnings = FALSE)
+
+########################################################################################
+#
+# Cnetplot
+#
+########################################################################################
+GO_results <- list()
+methods <- c('causarray', 'ruv')
+for (method in methods) {
+    df_list <- list()
+    for (dataset in c('ROSMAP-AD/results', 'SEA-AD/results-MTG', 'SEA-AD/results-PFC')){
+        path_rs <- sprintf('~/../jinandmaya/%s/%s/DE/', dataset, ps_model)
+        df <- read.csv(sprintf('%sres.%s.%s.csv', path_rs, celltype, method), row.names=1)
+        if('tau' %in% colnames(df)){df$log2FoldChange <- df$tau / log(2)}
+        df_list <- append(df_list, list(df))
+    }
+    # Get average log fold change for each gene
+    foldChange <- sapply(df$gene_names, function(gene) {
+        logfoldchanges <- sapply(df_list, function(df) {return(df$log2FoldChange[df$gene_names == gene])})
+        median(logfoldchanges, na.rm = TRUE)
+    })
+    
+    # Get overlap of discoveries
+    significant_genes <- unique(unlist(lapply(df_list, function(df) df$gene_names[df$padj <= 0.1])))
+    significant_genes_logfoldchange <- foldChange[significant_genes]
+    
+    # Select top 50 genes based on absolute log fold change
+    genes_to_test <- c(names(sort(significant_genes_logfoldchange, decreasing = TRUE))[1:25],
+        names(sort(significant_genes_logfoldchange, decreasing = FALSE))[1:25])
+    # genes_to_test <- names(significant_genes_logfoldchange)
+    length(genes_to_test)
+    
+    
+    #Run groupGO using the names of the vector
+    GO_results[[method]] <- groupGO(
+           gene     = genes_to_test,
+           OrgDb    = org.Hs.eg.db, keyType = "SYMBOL",
+           ont      = 'BP',
+           level    = 3,
+           readable = TRUE)
+    
+    GO_results[[method]]@result <- GO_results[[method]]@result[GO_results[[method]]@result$Count>0,]
+    GO_results[[method]]@result <- GO_results[[method]]@result[order(-GO_results[[method]]@result$Count),]
+    write.csv(GO_results[[method]]@result, sprintf("%sGO_%s.csv", path_rs_GO, method))
+    
+    #Plot your results
+    p <- cnetplot(GO_results[[method]], showCategory=GO_results[[method]]@result$Description[1:5],
+             categorySize="GeneNum",
+             circular = TRUE, colorEdge = TRUE, cex_label_gene = 1,
+             color.params = list(foldChange = foldChange), color_category = "#309630",
+             order=TRUE) + theme(plot.margin = unit(c(0, 0, 0, 0), "cm"))
+    
+    ggsave(sprintf("%scnetplot_%s.pdf", path_rs_GO, method), plot = p, width = 10, height = 8, units = "in")
+}

paper/AD/Plot.ipynb

@@ -0,0 +1,112 @@
+"","ID","Description","Count","GeneRatio","geneID"
+"GO:0050794","GO:0050794","regulation of cellular process",18,"18/50","SCGN/PDYN/GABRQ/COL6A2/MGARP/NGF/FAM83B/FKBP1C/OR7D2/PSG1/SELP/PSG6/VAX2/CARTPT/TRPA1/RAG2/RFPL1/PDC"
+"GO:0050789","GO:0050789","regulation of biological process",18,"18/50","SCGN/PDYN/GABRQ/COL6A2/MGARP/NGF/FAM83B/FKBP1C/OR7D2/PSG1/SELP/PSG6/VAX2/CARTPT/TRPA1/RAG2/RFPL1/PDC"
+"GO:0007154","GO:0007154","cell communication",15,"15/50","SCGN/PDYN/GABRQ/COL6A2/NGF/FAM83B/OR7D2/PSG1/SELP/PSG6/VAX2/CARTPT/TRPA1/RFPL1/PDC"
+"GO:0051716","GO:0051716","cellular response to stimulus",15,"15/50","PDYN/GABRQ/COL6A2/MGARP/NGF/FAM83B/OR7D2/PSG1/SELP/PSG6/VAX2/CARTPT/TRPA1/RFPL1/PDC"
+"GO:0023052","GO:0023052","signaling",15,"15/50","SCGN/PDYN/GABRQ/COL6A2/NGF/FAM83B/OR7D2/PSG1/SELP/PSG6/VAX2/CARTPT/TRPA1/RFPL1/PDC"
+"GO:0007165","GO:0007165","signal transduction",14,"14/50","PDYN/GABRQ/COL6A2/NGF/FAM83B/OR7D2/PSG1/SELP/PSG6/VAX2/CARTPT/TRPA1/RFPL1/PDC"
+"GO:0006807","GO:0006807","nitrogen compound metabolic process",10,"10/50","HGFAC/LDHC/NGF/FKBP1C/OR7D2/VAX2/CARTPT/RAG2/RFPL1/ADAM21"
+"GO:0044238","GO:0044238","primary metabolic process",10,"10/50","HGFAC/LDHC/NGF/FKBP1C/OR7D2/VAX2/CARTPT/RAG2/RFPL1/ADAM21"
+"GO:0071704","GO:0071704","organic substance metabolic process",10,"10/50","HGFAC/LDHC/NGF/FKBP1C/OR7D2/VAX2/CARTPT/RAG2/RFPL1/ADAM21"
+"GO:0044237","GO:0044237","cellular metabolic process",9,"9/50","HGFAC/LDHC/CA12/NGF/OR7D2/VAX2/CARTPT/RAG2/RFPL1"
+"GO:0009058","GO:0009058","biosynthetic process",8,"8/50","HGFAC/LDHC/NGF/OR7D2/VAX2/CARTPT/RAG2/RFPL1"
+"GO:0003008","GO:0003008","system process",8,"8/50","PDYN/GABRQ/NGF/OR7D2/VAX2/CARTPT/TRPA1/PDC"
+"GO:0048518","GO:0048518","positive regulation of biological process",8,"8/50","SCGN/MGARP/NGF/FAM83B/SELP/CARTPT/RAG2/RFPL1"
+"GO:0006950","GO:0006950","response to stress",8,"8/50","CLEC2A/HGFAC/MGARP/SELP/CARTPT/TRPA1/RAG2/RFPL1"
+"GO:0048522","GO:0048522","positive regulation of cellular process",7,"7/50","SCGN/MGARP/NGF/FAM83B/SELP/CARTPT/RFPL1"
+"GO:0048523","GO:0048523","negative regulation of cellular process",7,"7/50","MGARP/NGF/FKBP1C/VAX2/CARTPT/RAG2/RFPL1"
+"GO:0023051","GO:0023051","regulation of signaling",7,"7/50","SCGN/NGF/FAM83B/SELP/CARTPT/RFPL1/PDC"
+"GO:0048519","GO:0048519","negative regulation of biological process",7,"7/50","MGARP/NGF/FKBP1C/VAX2/CARTPT/RAG2/RFPL1"
+"GO:0009605","GO:0009605","response to external stimulus",7,"7/50","CLEC2A/SELP/CARTPT/TRPA1/RAG2/RFPL1/PDC"
+"GO:0022414","GO:0022414","reproductive process",6,"6/50","LDHC/CA12/PSG1/PSG6/ADAM21/PSG11"
+"GO:0019222","GO:0019222","regulation of metabolic process",6,"6/50","NGF/OR7D2/VAX2/CARTPT/RAG2/RFPL1"
+"GO:0071840","GO:0071840","cellular component organization or biogenesis",6,"6/50","MGARP/NGF/SELP/VAX2/TRPA1/RAG2"
+"GO:0007275","GO:0007275","multicellular organism development",6,"6/50","MGARP/NGF/VAX2/CARTPT/RAG2/ADAM21"
+"GO:0048856","GO:0048856","anatomical structure development",6,"6/50","MGARP/NGF/VAX2/CARTPT/RAG2/ADAM21"
+"GO:0023056","GO:0023056","positive regulation of signaling",6,"6/50","SCGN/NGF/FAM83B/SELP/CARTPT/RFPL1"
+"GO:0048583","GO:0048583","regulation of response to stimulus",6,"6/50","NGF/FAM83B/SELP/CARTPT/RFPL1/PDC"
+"GO:0042221","GO:0042221","response to chemical",6,"6/50","COL6A2/MGARP/OR7D2/SELP/CARTPT/TRPA1"
+"GO:0051234","GO:0051234","establishment of localization",6,"6/50","GABRQ/MGARP/FKBP1C/CLEC4F/CARTPT/TRPA1"
+"GO:0065008","GO:0065008","regulation of biological quality",6,"6/50","SCGN/HGFAC/GABRQ/NGF/SELP/CARTPT"
+"GO:0002682","GO:0002682","regulation of immune system process",5,"5/50","PSG1/SELP/PSG6/CARTPT/RAG2"
+"GO:0048869","GO:0048869","cellular developmental process",5,"5/50","MGARP/NGF/VAX2/CARTPT/RAG2"
+"GO:0048878","GO:0048878","chemical homeostasis",5,"5/50","SCGN/CA12/FKBP1C/CARTPT/TRPA1"
+"GO:0048584","GO:0048584","positive regulation of response to stimulus",5,"5/50","NGF/FAM83B/SELP/CARTPT/RFPL1"
+"GO:0007155","GO:0007155","cell adhesion",4,"4/50","COL6A2/SELP/RAG2/PSG11"
+"GO:0008283","GO:0008283","cell population proliferation",4,"4/50","NGF/FAM83B/RAG2/RFPL1"
+"GO:0051641","GO:0051641","cellular localization",4,"4/50","MGARP/FKBP1C/CARTPT/RFPL1"
+"GO:0051239","GO:0051239","regulation of multicellular organismal process",4,"4/50","NGF/SELP/CARTPT/RAG2"
+"GO:0051240","GO:0051240","positive regulation of multicellular organismal process",4,"4/50","NGF/SELP/CARTPT/RAG2"
+"GO:0050793","GO:0050793","regulation of developmental process",4,"4/50","MGARP/NGF/CARTPT/RAG2"
+"GO:0019725","GO:0019725","cellular homeostasis",4,"4/50","SCGN/FKBP1C/CARTPT/TRPA1"
+"GO:0051707","GO:0051707","response to other organism",4,"4/50","CLEC2A/SELP/RAG2/RFPL1"
+"GO:0009607","GO:0009607","response to biotic stimulus",4,"4/50","CLEC2A/SELP/RAG2/RFPL1"
+"GO:0009628","GO:0009628","response to abiotic stimulus",4,"4/50","COL6A2/MGARP/TRPA1/PDC"
+"GO:0065009","GO:0065009","regulation of molecular function",4,"4/50","NGF/FKBP1C/CARTPT/RFPL1"
+"GO:0006955","GO:0006955","immune response",3,"3/50","CLEC2A/RAG2/RFPL1"
+"GO:0009893","GO:0009893","positive regulation of metabolic process",3,"3/50","NGF/CARTPT/RFPL1"
+"GO:0008219","GO:0008219","cell death",3,"3/50","COL6A2/NGF/RFPL1"
+"GO:0055085","GO:0055085","transmembrane transport",3,"3/50","GABRQ/FKBP1C/TRPA1"
+"GO:0044703","GO:0044703","multi-organism reproductive process",3,"3/50","PSG1/PSG6/PSG11"
+"GO:0044706","GO:0044706","multi-multicellular organism process",3,"3/50","PSG1/PSG6/PSG11"
+"GO:0051093","GO:0051093","negative regulation of developmental process",3,"3/50","MGARP/CARTPT/RAG2"
+"GO:0032879","GO:0032879","regulation of localization",3,"3/50","FKBP1C/CARTPT/RFPL1"
+"GO:0051606","GO:0051606","detection of stimulus",3,"3/50","OR7D2/TRPA1/PDC"
+"GO:0033036","GO:0033036","macromolecule localization",3,"3/50","MGARP/CARTPT/RFPL1"
+"GO:0002252","GO:0002252","immune effector process",2,"2/50","CLEC2A/RAG2"
+"GO:0002683","GO:0002683","negative regulation of immune system process",2,"2/50","CARTPT/RAG2"
+"GO:0009056","GO:0009056","catabolic process",2,"2/50","LDHC/RFPL1"
+"GO:0044281","GO:0044281","small molecule metabolic process",2,"2/50","LDHC/CA12"
+"GO:0001775","GO:0001775","cell activation",2,"2/50","SELP/RAG2"
+"GO:0007017","GO:0007017","microtubule-based process",2,"2/50","LDHC/MGARP"
+"GO:0048870","GO:0048870","cell motility",2,"2/50","LDHC/SELP"
+"GO:0007610","GO:0007610","behavior",2,"2/50","NGF/CARTPT"
+"GO:0050817","GO:0050817","coagulation",2,"2/50","HGFAC/SELP"
+"GO:0051241","GO:0051241","negative regulation of multicellular organismal process",2,"2/50","CARTPT/RAG2"
+"GO:0009653","GO:0009653","anatomical structure morphogenesis",2,"2/50","NGF/VAX2"
+"GO:0048589","GO:0048589","developmental growth",2,"2/50","NGF/RAG2"
+"GO:0051094","GO:0051094","positive regulation of developmental process",2,"2/50","NGF/RAG2"
+"GO:0040008","GO:0040008","regulation of growth",2,"2/50","NGF/RAG2"
+"GO:0045927","GO:0045927","positive regulation of growth",2,"2/50","NGF/RAG2"
+"GO:0009719","GO:0009719","response to endogenous stimulus",2,"2/50","MGARP/NGF"
+"GO:0019953","GO:0019953","sexual reproduction",1,"1/50","ADAM21"
+"GO:0032504","GO:0032504","multicellular organism reproduction",1,"1/50","CA12"
+"GO:0002200","GO:0002200","somatic diversification of immune receptors",1,"1/50","RAG2"
+"GO:0002520","GO:0002520","immune system development",1,"1/50","RAG2"
+"GO:0002684","GO:0002684","positive regulation of immune system process",1,"1/50","SELP"
+"GO:0045321","GO:0045321","leukocyte activation",1,"1/50","RAG2"
+"GO:0050900","GO:0050900","leukocyte migration",1,"1/50","SELP"
+"GO:0009892","GO:0009892","negative regulation of metabolic process",1,"1/50","VAX2"
+"GO:0001906","GO:0001906","cell killing",1,"1/50","CLEC2A"
+"GO:0007049","GO:0007049","cell cycle",1,"1/50","RFPL1"
+"GO:0016049","GO:0016049","cell growth",1,"1/50","NGF"
+"GO:0016192","GO:0016192","vesicle-mediated transport",1,"1/50","CLEC4F"
+"GO:0022402","GO:0022402","cell cycle process",1,"1/50","RFPL1"
+"GO:0051301","GO:0051301","cell division",1,"1/50","RFPL1"
+"GO:0051651","GO:0051651","maintenance of location in cell",1,"1/50","FKBP1C"
+"GO:0061919","GO:0061919","process utilizing autophagic mechanism",1,"1/50","RFPL1"
+"GO:0140352","GO:0140352","export from cell",1,"1/50","CARTPT"
+"GO:0019058","GO:0019058","viral life cycle",1,"1/50","RFPL1"
+"GO:0050792","GO:0050792","regulation of viral process",1,"1/50","RFPL1"
+"GO:0003006","GO:0003006","developmental process involved in reproduction",1,"1/50","ADAM21"
+"GO:0009566","GO:0009566","fertilization",1,"1/50","ADAM21"
+"GO:0048609","GO:0048609","multicellular organismal reproductive process",1,"1/50","CA12"
+"GO:0097722","GO:0097722","sperm motility",1,"1/50","LDHC"
+"GO:0007389","GO:0007389","pattern specification process",1,"1/50","VAX2"
+"GO:0033555","GO:0033555","multicellular organismal response to stress",1,"1/50","TRPA1"
+"GO:0035265","GO:0035265","organ growth",1,"1/50","RAG2"
+"GO:0048771","GO:0048771","tissue remodeling",1,"1/50","CARTPT"
+"GO:0048871","GO:0048871","multicellular organismal-level homeostasis",1,"1/50","CARTPT"
+"GO:0040012","GO:0040012","regulation of locomotion",1,"1/50","SELP"
+"GO:0040017","GO:0040017","positive regulation of locomotion",1,"1/50","SELP"
+"GO:0002358","GO:0002358","B cell homeostatic proliferation",1,"1/50","RAG2"
+"GO:0044403","GO:0044403","biological process involved in symbiotic interaction",1,"1/50","RFPL1"
+"GO:0007623","GO:0007623","circadian rhythm",1,"1/50","CARTPT"
+"GO:0042698","GO:0042698","ovulation cycle",1,"1/50","CA12"
+"GO:0051050","GO:0051050","positive regulation of transport",1,"1/50","CARTPT"
+"GO:0023057","GO:0023057","negative regulation of signaling",1,"1/50","CARTPT"
+"GO:0048585","GO:0048585","negative regulation of response to stimulus",1,"1/50","CARTPT"
+"GO:0051051","GO:0051051","negative regulation of transport",1,"1/50","CARTPT"
+"GO:0043903","GO:0043903","regulation of biological process involved in symbiotic interaction",1,"1/50","RFPL1"
+"GO:0051235","GO:0051235","maintenance of location",1,"1/50","FKBP1C"
+"GO:0051640","GO:0051640","organelle localization",1,"1/50","MGARP"