OSCR

Uncovering Latent Structure in Gliomas Using Multi-Omics Factor Analysis.

Code ↔ Paper

4 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 4 matches
  1. [1] § 2. Materials and Methods › 2.3. Multi-Omics Factor Analysis (MOFA) ↔ run_mofa.R, lines 18–94 · score 0.74 · MOFA2, MOFA model, medium, training, seed, likelihoods
  2. [2] § 2. Materials and Methods › 2.5. Gene Set Enrichment Analysis (GSEA) ↔ enrichment.R, lines 50–99 · score 0.73 · stExon, TSS1500, TSS200, bias, BP, GO
  3. [3] § 3. Results › 3.1. Overview of the Model › 3.1.1. Features Selected ↔ mofa-results.R, lines 351–433 · score 0.60 · gene ISM1, probes selected, mRNA, CpGs, DNA, matrices
  4. [4] § 3. Results › 3.2. Interpretation and Evaluation of the Findings ↔ mofa-results.R, lines 351–433 · score 0.54 · Pearson correlations, gene regulation, ISM1, probes

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

R · 508 lines · 18 KB · no license · 2 matches

  1. setwd("~/Desktop/artigo")
  3. source("setup.R")
  4. ### Load data
  5. clinical <- read.csv("DataSets/2_analysis/clinical.csv", row.names = 1)
  7. survival <- read.csv("DataSets/2_analysis/survival_data.csv", row.names = 1)
  9. mutations <- read.csv("DataSets/2_analysis/mutations.csv", row.names = 1)
  10. dna <- read.csv("DataSets/2_analysis/dna.csv", row.names = 1)
  11. mrna <- read.csv("DataSets/2_analysis/mrna.csv", row.names = 1)
  12. mirna <- read.csv("DataSets/2_analysis/mirna.csv", row.names = 1)
  15. info.methy <- read.csv("DataSets/processed_assays/info_methylation.csv", row.names = 1)
  16. info.mirna <- read.csv("DataSets/processed_assays/info_rna_mirna.csv", row.names = 1)
  17. info.mrna <- read.csv("DataSets/processed_assays/info_rna_coding.csv", row.names = 1)
  19. omics.list <- list(Mutations = mutations, Methylation = dna, mRNA = mrna, miRNA = mirna)
  24. ## Run Model
  25. source("run_mofa.R")
  27. output <- run_mofa(omics.list, n_runs = 10, num_factors = NULL,
  28. likelihoods = c("bernoulli", "gaussian", "gaussian", "gaussian"),
  29. clinical_data = clinical)
  33. #### Save Results
  34. model <- output$model_output
  35. print(model)
  36. print(output$time_taken)
  40. #### Read Z and W
  41. factors <- as.matrix(read.csv("mofa-output/factors.csv", row.names = 1))
  43. loadings <- list(
  44. Mutations = as.matrix(read.csv("mofa-output/loadings_mutations.csv", row.names = 1)),
  45. Methylation = as.matrix(read.csv("mofa-output/loadings_Methylation.csv", row.names = 1)),
  46. mRNA = as.matrix(read.csv("mofa-output/loadings_mrna.csv", row.names = 1)),
  47. miRNA = as.matrix(read.csv("mofa-output/loadings_mirna.csv", row.names = 1))
  48. )
  50. ### Results shown in the paper
  52. ## Variance Decomposition
  54. p__variance_decomposition <- local({
  55. p1 <- plot_variance_explained(model, plot_total = T)[[1]]
  56. p1 <- p1$data
  58. p1$value <- round(p1$value, 1)
  59. colnames(p1) <- c("Factor", "View", "Value", "Group")
  60. p1$Factor <- gsub("([a-zA-Z]+)([0-9]+)", "\\1 \\2", p1$Factor)
  61. p1$View <- factor(p1$View, levels = c("miRNA", "mRNA", "Methylation", "Mutations"))
  63. p <- ggplot(p1, aes(x = .data$Value, y = .data$View, fill = .data$Factor)) +
  64. geom_bar(
  65. position = position_stack(reverse = T), stat = "identity") +
  66. stat_identity(
  67. geom = "text", color = "black", size = 5,
  68. aes(label = ifelse(.data$Value > 2, .data$Value, "")),
  69. position = position_stack(reverse = TRUE, vjust = 0.5)
  70. ) +
  71. labs(title = "", x = "% of Variance", y = "") +
  72. scale_fill_manual(values = brewer.pal(n = 4, name = "Oranges")) +
  73. my_theme
  75. p
  76. })
  77. ggsave(filename = "mofa-results/variance-decomposition.pdf",
  78. p__variance_decomposition, width = 14, height = 7)
  82. ## Factors projection
  83. classified_samples <- clinical[!is.na(clinical$Subtype), ]
  85. p__factors_projection <- local({
  86. sample_df <- data.frame(Factor1 = numeric(nrow(classified_samples)),
  87. Factor3 = numeric(nrow(classified_samples)),
  88. Factor2 = numeric(nrow(classified_samples)),
  89. Subtype = rep("GBM", nrow(classified_samples)))
  90. sample_df$Factor1 <- factors[rownames(classified_samples),1]
  91. sample_df$Factor3 <- factors[rownames(classified_samples),3]
  92. sample_df$Factor2 <- factors[rownames(classified_samples),2]
  93. sample_df$Subtype <- classified_samples$Subtype
  95. p <- ggplot(sample_df, aes(x = Factor1, y = Factor3)) +
  96. geom_point(aes(shape = Subtype, color = Factor2), size = 2.5) +
  97. geom_vline(xintercept=0, linetype="dashed") +
  98. geom_segment(aes(x = 0, xend = Inf, y = 0, yend = 0), linetype = "dashed") +
  99. scale_color_gradientn(
  100. colours = c("darkgreen", "white", "darkred"),
  101. name = "Factor 2"
  102. ) +
  103. my_theme +
  104. labs(x = "Factor 1", y = "Factor 3", shape = "Subtype")
  105. p
  106. })
  107. ggsave(filename = "mofa-results/factors-projectionUPDATED.pdf", p__factors_projection, width = 6, height = 4)
  111. ## Hazard Ratio for factors
  112. plot_survival_factors_univariate <- function(survival, factors_matrix) {
  114. survival[is.na(survival)] <- 0
  115. surv_object <- Surv(time = survival[rownames(factors_matrix), "Time"],
  116. event = survival[rownames(factors_matrix), "Status"])
  118. results <- lapply(colnames(factors_matrix), function(fac) {
  119. factor_values <- factors_matrix[, fac, drop = FALSE]
  120. fit <- coxph(surv_object ~ factor_values)
  121. s <- summary(fit)
  122. coef <- s$coefficients
  123. conf <- s$conf.int
  124. data.frame(
  125. factor = fac,
  126. coef = coef["factor_values", "exp(coef)"],
  127. p = coef["factor_values", "Pr(>|z|)"],
  128. lower = conf["factor_values", "lower .95"],
  129. higher = conf["factor_values", "upper .95"]
  130. )
  131. })
  133. df <- do.call(rbind, results)
  134. df$factor <- factor(df$factor, levels = rev(df$factor))
  135. df$significant <- ifelse(df$p < 0.05, "*", "")
  137. p <- ggplot(df, aes(x = factor, y = coef, ymin = lower, ymax = higher)) +
  138. geom_pointrange() +
  139. geom_text(aes(label = significant, y = coef), vjust = -1, size = 5)+
  140. labs(y="Hazard Ratio", x="") +
  141. scale_x_discrete(labels = paste0("Factor ", seq(ncol(factors_matrix),1))) +
  142. geom_hline(aes(yintercept=1), linetype="dotted") +
  143. coord_flip() +
  144. my_theme
  146. return(list(p, df))
  147. }
  149. plot_survival_factors_multivariate <- function (survival, factors_matrix){
  151. survival[is.na(survival)] <- 0
  152. surv_object <- Surv(time = survival[rownames(factors_matrix), "Time"], event = survival[rownames(factors_matrix), "Status"] )
  154. fit <- coxph(surv_object ~ factors_matrix)
  156. s <- summary(fit)
  157. coef <- s[["coefficients"]]
  159. df <- data.frame(
  160. factor = factor(rownames(coef), levels = rev(rownames(coef))),
  161. p = coef[,"Pr(>|z|)"],
  162. coef = coef[,"exp(coef)"],
  163. lower = s[["conf.int"]][,"lower .95"],
  164. higher = s[["conf.int"]][,"upper .95"])
  166. df$significant <- ifelse(df$p < 0.05, "*", "")
  168. p <- ggplot(df, aes(x = factor, y = coef, ymin = lower, ymax = higher)) +
  169. geom_pointrange() +
  170. geom_text(aes(label = significant, y = coef), vjust = -1, size = 5)+
  171. labs(y="Hazard Ratio", x="") +
  172. scale_x_discrete(labels = paste0("Factor ", seq(ncol(factors_matrix),1))) +
  173. geom_hline(aes(yintercept=1), linetype="dotted") +
  174. coord_flip() +
  175. my_theme
  176. print(p)
  178. return(list(p, fit))
  179. }
  181. p <- plot_survival_factors_univariate(survival, factors)[[1]]
  182. ggsave(filename = "mofa-results/factors-hazard-uni.pdf", p, width = 6, height = 4)
  184. p <- plot_survival_factors_multivariate(survival, factors)[[1]]
  185. ggsave(filename = "mofa-results/factors-hazard-multi.pdf", p, width = 6, height = 4)
  187. rm(p)
  190. ## Features selected & mutations projections
  192. # Mutations Omic
  193. mutations_sel_f1_pos <- c("IDH1")
  194. mutations_sel_f1_neg <- c("PTEN", "EGFR")
  196. mutations_sel_f3_pos <- c("TP53", "ATRX")
  197. mutations_sel_f3_neg <- c("CIC")
  199. p__mutations_projection <- local({
  200. combs <- c("EGFR", "IDH1", "IDH1", "ATRX", "CIC", "TP53")
  201. p_list <- list()
  202. for (k in c(1, 3, 5)){
  203. p <- plot_factors(model,
  204. factors = c(1,3),
  205. color_by = combs[k],
  206. shape_by = combs[k+1],
  207. show_missing = T,
  208. scale = T
  209. )
  210. p <- p +
  211. geom_vline(xintercept=0, linetype="dashed") +
  212. geom_segment(aes(x = 0, xend = max(p$data$x), y = 0, yend = 0), linetype = "dashed") +
  213. my_theme +
  214. labs(x = "Factor 1", y = "Factor 3")
  216. ggsave(filename = paste0("mofa-results/mutations-projected-", k, ".pdf"), p, width = 6, height = 4)
  217. p_list <- append(p_list, p)
  218. }
  220. p_list
  221. })
  224. # DNA Methylation Omic
  225. methy_sel_f1 <- names(head(sort(abs(loadings$Methylation[, 1]), decreasing = TRUE), 30))
  226. methy_sel_f3 <- names(head(sort(abs(loadings$Methylation[, 3]), decreasing = TRUE), 30))
  228. # mRNA Omic
  229. mrna_sel_f1 <- names(head(sort(abs(loadings$mRNA[, 1]), decreasing = TRUE), 30))
  230. mrna_sel_f2 <- names(head(sort(abs(loadings$mRNA[, 2]), decreasing = TRUE), 30))
  231. mrna_sel_f3 <- names(head(sort(abs(loadings$mRNA[, 3]), decreasing = TRUE), 30))
  233. # miRNA Omic
  234. mirna_sel_f1 <- names(head(sort(abs(loadings$miRNA[, 1]), decreasing = TRUE), 3))
  238. ## Enrichment Analysis
  239. source("enrichment.R")
  241. # mRNA
  242. mrna_enrich_f1 <- run_rna_gsea(loadings$mRNA[, 1], sign = 0, alpha = 0.05)
  243. plot_gsea(mrna_enrich_f1, alpha = 0.05, max.pathways = 25, sign = -2, filename = NULL)
  245. mrna_enrich_f2 <- run_rna_gsea(loadings$mRNA[, 2], sign = 0, alpha = 0.05)
  246. plot_gsea(mrna_enrich_f2, alpha = 0.05, max.pathways = 25, sign = 2, filename = NULL)
  248. mrna_enrich_f3 <- run_rna_gsea(loadings$mRNA[, 3], sign = 0, alpha = 0.05)
  249. plot_gsea(mrna_enrich_f3, alpha = 0.05, max.pathways = 25, sign = 2, filename = NULL)
  251. plot_categories_reactome(list("Factor 1"= mrna_enrich_f1[mrna_enrich_f1$NES<0,"ID"],
  252. "Factor 2"= mrna_enrich_f2[mrna_enrich_f2$NES>0, "ID"],
  253. "Factor 3"= mrna_enrich_f3[mrna_enrich_f3$NES>0, "ID"]),
  254. filename="mrna-gsea-top-levels")
  255. #the pathways in "Disease"
  256. diseases_f1 <- pathways_of_disease(mrna_enrich_f1)
  257. print(table(diseases_f1))
  259. diseases_f2 <- pathways_of_disease(mrna_enrich_f2)
  260. print(table(diseases_f2))
  262. diseases_f3 <- pathways_of_disease(mrna_enrich_f3)
  263. print(table(diseases_f3))
  266. # DNA methylation
  267. methy_enrich_f1 <- run_methy_gsea(loadings$Methylation[,1], sign = "positive",
  268. background_vector = rownames(loadings$Methylation),
  269. q = 0.01, promoter = F, alpha = 0.05, filename = NULL)
  270. plot_methy_gsea(methy_enrich_f1, sign = 2, max.pathways = 15,
  271. filename = "methy-enrich-f1",
  272. alpha = 0.05)
  274. methy_enrich_f1_promoter <- run_methy_gsea(loadings$Methylation[,1], sign = "positive",
  275. background_vector = rownames(loadings$Methylation), q = 0.01,
  276. promoter = TRUE, alpha = 0.05, filename = NULL)
  278. plot_methy_gsea(methy_enrich_f1_promoter, sign = 2, max.pathways = 15,
  279. filename = "methy-enrich-f1-prom",
  280. alpha = 0.05)
  284. methy_enrich_f3 <- run_methy_gsea(loadings$Methylation[,3], sign = "negative",
  285. background_vector = rownames(loadings$Methylation),
  286. q = 0.01, promoter = F, alpha = 0.05, filename = NULL)
  287. plot_methy_gsea(methy_enrich_f3, sign = -2, max.pathways = 15,
  288. filename = "methy-enrich-f3",
  289. alpha = 0.05)
  292. methy_enrich_f3_promoter <- run_methy_gsea(loadings$Methylation[,3], sign = "negative",
  293. background_vector = rownames(loadings$Methylation),
  294. q = 0.01,
  295. promoter = TRUE, alpha = 0.05, filename = NULL)
  296. plot_methy_gsea(methy_enrich_f3_promoter, sign = -2, max.pathways = 15,
  297. filename = "methy-enrich-f3-prom",
  298. alpha = 0.05)
  304. ## Validation Features
  307. # Survival
  308. gbm_mrna <- read.csv("survival-results/gbm_mrna.csv", row.names = 1)
  309. lgg_mrna <- read.csv("survival-results/lgg_mrna.csv", row.names = 1)
  311. gbm_methy <- read.csv("survival-results/gbm_methy.csv", row.names = 1)
  312. lgg_methy <- read.csv("survival-results/lgg_methy.csv", row.names = 1)
  316. # DGE
  317. gbm_vs_astro_mrna <- read.csv("dge-results/gbm_vs_astro_mrna.csv", row.names = 1)
  318. gbm_vs_oligo_mrna <- read.csv("dge-results/gbm_vs_oligo_mrna.csv", row.names = 1)
  319. astro_vs_oligo_mrna <- read.csv("dge-results/astro_vs_oligo_mrna.csv", row.names = 1)
  321. gbm_vs_astro_methy <- read.csv("dge-results/gbm_vs_astro_methy.csv", row.names = 1)
  322. gbm_vs_oligo_methy <- read.csv("dge-results/oligo_vs_gbm_methy.csv", row.names = 1)
  323. astro_vs_oligo_methy <- read.csv("dge-results/astro_vs_oligo_methy.csv", row.names = 1)
  325. # build data frame as summary for each factor:
  326. #DGE (GBM vs ASTRO; GBM vs OLIGO; ASTRO vs OLIGO);
  327. #Significant in survival: in GBM group || in LGG group
  328. build_df <- function(methy_sel, mrna_sel) {
  330. combined_features <- c(methy_sel, mrna_sel)
  331. df <- data.frame(matrix(NA, nrow = length(combined_features), ncol = 5))
  332. colnames(df) <- c("GBMvsASTRO", "GBMvsOLIGO", "ASTROvsOLIGO", "GBM", "LGG")
  333. rownames(df) <- combined_features
  335. df$"GBMvsASTRO" <- c(gbm_vs_astro_methy[methy_sel, "diffexpressed"],
  336. gbm_vs_astro_mrna[mrna_sel, "diffexpressed"])
  338. df$"GBMvsOLIGO" <- c(gbm_vs_oligo_methy[methy_sel, "diffexpressed"],
  339. gbm_vs_oligo_mrna[mrna_sel, "diffexpressed"])
  341. df$"ASTROvsOLIGO" <- c(astro_vs_oligo_methy[methy_sel, "diffexpressed"],
  342. astro_vs_oligo_mrna[mrna_sel, "diffexpressed"])
  344. df$"LGG" <- round(c(lgg_methy[methy_sel, "P_Value"],
  345. lgg_mrna[mrna_sel, "P_Value"]),3)
  347. df$"GBM" <- round(c(gbm_methy[methy_sel, "P_Value"],
  348. gbm_mrna[mrna_sel, "P_Value"]),3)
  349. return(df)
  350. }
  351. df_f1 <- build_df(methy_sel_f1, mrna_sel_f1)
  352. df_f2 <- build_df(character(0), mrna_sel_f2)
  353. df_f3 <- build_df(methy_sel_f3, mrna_sel_f3)
  355. write.csv(df_f1,"summary-f1.csv", row.names = TRUE)
  356. write.csv(df_f2,"summary-f2.csv", row.names = TRUE)
  357. write.csv(df_f3,"summary-f3.csv", row.names = TRUE)
  361. ## Correlation of CpG and mRNA
  362. plot_cor_cpgs_mrna <- local({
  363. conc_matrix <- cbind(omics.list$Methylation, omics.list$mRNA)
  364. conc_matrix <- conc_matrix[,c(methy_sel_f1, mrna_sel_f1, methy_sel_f3, mrna_sel_f3)]
  365. correlation_matrix <- cor(conc_matrix, method = "pearson", use = "na.or.complete")
  366. rownames(correlation_matrix) <- c(methy_sel_f1,
  367. info.mrna[rownames(info.mrna) %in% mrna_sel_f1, "gene_name"],
  368. methy_sel_f3,
  369. info.mrna[rownames(info.mrna) %in%mrna_sel_f3, "gene_name"])
  371. col_fun <- colorRamp2(c(-1, 0, 1), c("#268989", "white", "#E43F3F"))
  373. ht <- Heatmap(
  374. correlation_matrix,
  375. name = "Pearson Correlation",
  376. col = col_fun,
  377. cluster_rows = FALSE,
  378. cluster_columns = FALSE,
  379. show_row_names = TRUE,
  380. show_column_names = FALSE,
  381. row_names_gp = gpar(fontsize = 20),
  382. heatmap_legend_param = list(
  383. title = NULL,
  384. title_gp = gpar(fontsize = 0),
  385. labels_gp = gpar(fontsize = 20),
  386. legend_height = unit(75, "cm")
  387. )
  388. )
  389. pdf("mofa-results/heatmap-cpgs-mrna.pdf", width = 30, height = 30)
  390. draw(ht)
  391. dev.off()
  392. })
  396. ## Cpg-Gene regulation
  397. #in particular, gene ISM1 with the 8 probes selected
  398. cpgs_f3_ism1 <- intersect(methy_sel_f3, rownames(info.methy)[info.methy$gene=="ISM1"])
  399. id_ism1 <- rownames(info.mrna)[info.mrna$gene_name=="ISM1"]
  400. expr_ism1 <- mrna[, id_ism1, drop = FALSE]
  402. lgg_patients <- rownames(clinical)[clinical$Type == "LGG" & !is.na(clinical$Type)]
  403. gbm_patients <- rownames(clinical)[clinical$Type == "GBM" & !is.na(clinical$Type)]
  405. # correlations of each cpg with the gene ISM1
  406. cors <- local({
  407. cors <- c()
  408. for (cpg in cpgs_f3_ism1){
  409. val <- cor(expr_ism1[lgg_patients, ], dna[lgg_patients, cpg], method = "pearson", use = "complete.obs")
  410. cors <- append(cors, val)
  411. }
  412. cors
  413. })
  414. cors
  416. # correlations of each cpg with the genes selected by factor 3 (choose idx)
  417. cors_matrix <- local({
  418. cors_matrix <- matrix(data = NA,
  419. nrow = length(cpgs_f3_ism1),
  420. ncol = length(mrna_sel_f3),
  421. dimnames = list(cpgs_f3_ism1, mrna_sel_f3))
  422. for (idx_cpg in 1: length(cpgs_f3_ism1)){
  423. for (idx_gene in 1: length(mrna_sel_f3)){
  424. gene_name <- info.mrna[rownames(info.mrna) == mrna_sel_f3[idx_gene], "gene_name"]
  425. val <- cor(mrna[lgg_patients, mrna_sel_f3[idx_gene]], dna[lgg_patients, cpgs_f3_ism1[idx_cpg]],
  426. method = "pearson", use = "complete.obs")
  427. cors_matrix[idx_cpg, idx_gene] <- val
  428. }
  429. }
  430. colnames(cors_matrix) <- info.mrna[rownames(info.mrna) %in% mrna_sel_f3, "gene_name"]
  431. rownames(cors_matrix) <- cpgs_f3_ism1
  432. cors_matrix
  433. })
  436. # Plot
  437. cor_plot_gene_cpg <- function(cpg, gene_id, labels, groups_2_test){
  438. idxs <- which(labels %in% groups_2_test)
  439. expr_gene <- mrna[idxs , gene_id, drop = FALSE]
  440. gene_name <- info.mrna[rownames(info.mrna) == gene_id, "gene_name"]
  441. expr_cpg <- dna[idxs, cpg, drop = FALSE]
  443. df_plot <- data.frame(
  444. expr_gene = expr_gene[,1],
  445. expr_cpg = expr_cpg[,1],
  446. groups = labels[idxs]
  447. )
  449. plot <- ggplot(df_plot, aes(x = expr_cpg, y = expr_gene, color = groups)) +
  450. geom_smooth(method = "lm", se = T) +
  451. geom_point() +
  452. scale_color_manual(values = c('#FF7400', '#009999', "black")) +
  453. stat_cor(method = "pearson",
  454. cor.coef.name = "R",
  455. label.x = 1.08) +
  456. labs(
  457. x = cpg,,
  458. y = gene_name,
  459. color = "Subtype"
  460. ) +
  461. my_theme
  462. ggsave(filename = paste0("mofa-results/corr-",cpg, "-", gene_name,".pdf"), plot, width = 8, height = 6)
  463. return (plot)
  464. }
  466. p <- cor_plot_gene_cpg(cpgs_f3_ism1[1], id_ism1,
  467. labels = clinical$Subtype, groups_2_test = c("ASTRO", "OLIGO", "GBM"))
  468. p <- cor_plot_gene_cpg(cpgs_f3_ism1[2], id_ism1,
  469. labels = clinical$Subtype, groups_2_test = c("ASTRO", "OLIGO", "GBM"))
  471. # heatmap of cors_matrix
  472. col_fun <- colorRamp2(c(-1, 0, 1), c("#268989", "white", "#E43F3F"))
  474. cors_matrix <- cbind(cors_matrix, "ISM1" = cors)
  476. ht <- Heatmap(
  477. cors_matrix,
  478. name = "Pearson Correlation",
  479. col = col_fun,
  480. cluster_rows = FALSE,
  481. cluster_columns = FALSE,
  482. show_row_names = TRUE,
  483. show_column_names = T,
  484. row_names_gp = gpar(fontsize = 20),
  485. heatmap_legend_param = list(
  486. title = NULL,
  487. title_gp = gpar(fontsize = 0),
  488. labels_gp = gpar(fontsize = 20),
  489. legend_height = unit(35, "cm")
  490. )
  491. )
  492. pdf("mofa-results/heatmap-mrnaf3-cpgsiNism1.pdf", width = 20, height = 15)
  493. draw(ht)
  494. dev.off()
  497. # SLC2A5 BLNK TMEM119 PLXDC2
  498. id_plxdc2 <- rownames(info.mrna)[info.mrna$gene_name=="PLXDC2"]
  499. id_slc2a5 <- rownames(info.mrna)[info.mrna$gene_name=="SLC2A5"]
  500. id_blnk <- rownames(info.mrna)[info.mrna$gene_name=="BLNK"]
  501. id_tmem119 <- rownames(info.mrna)[info.mrna$gene_name=="TMEM119"]
  503. p <- cor_plot_gene_cpg(cpgs_f3_ism1[1], id_plxdc2,
  504. labels = clinical$Subtype, groups_2_test = c("ASTRO", "OLIGO", "GBM"))
  505. p <- cor_plot_gene_cpg(cpgs_f3_ism1[2], id_plxdc2,
  506. labels = clinical$Subtype, groups_2_test = c("ASTRO", "OLIGO", "GBM"))

mofa-results.R at commit 4c8df63, no license · at the source

Overview

Authors: Catarina Gameiro Carvalho1, Alexandra M. Carvalho2, Susana Vinga3,4
ORCID iDs: Alexandra M. Carvalho, Susana Vinga
  1. Instituto Superior Técnico, Universidade de Lisboa, 1049-001 Lisbon, Portugal
  2. Instituto de Telecomunicações, Instituto Superior Técnico, Universidade de Lisboa, 1049-001 Lisbon, Portugal
  3. Instituto de Engenharia de Sistemas e Computadores-Investigação e Desenvolvimento (INESC-ID), Instituto Superior Técnico, Universidade de Lisboa, 1000-029 Lisbon, Portugal
  4. IDMEC, Instituto Superior Técnico, Universidade de Lisboa, 1049-001 Lisbon, Portugal
Institutions: University of Lisbon (Portugal); Instituto Superior Técnico (Portugal); Instituto de Telecomunicações (Portugal); Instituto de Engenharia de Sistemas e Computadores Investigação e Desenvolvimento (Portugal); Instituto Superior Técnico Instituto de Engenharia Mecânica (Portugal)
Journal: Genes, volume 17, issue 5, article 540
Dates: received 13 March 2026; accepted 21 April 2026; published online 1 May 2026
Type: Research article · Language: English
License: CC BY
Identifiers: DOI 10.3390/genes17050540 · PMID 42194997 · PMCID PMC13205169 · OpenAlex W7160132433
Open access: gold, a free copy (OpenAlex)
Status: code verified
Categories: genetics / omics (modality), human (organism), other condition (population)
Methods: Statistics, Graphs
Keywords: multi-omics integration, prognostic biomarkers, survival analysis, molecular subtyping, latent factor model
MeSH: Brain Neoplasms*, Glioma*, Bayes Theorem, Biomarkers, Tumor, DNA Methylation, Factor Analysis, Statistical, Gene Expression Profiling, Gene Expression Regulation, Neoplastic, Humans, MicroRNAs, Multiomics, Prognosis, Transcriptome (* major topic)
Topic: Glioma Diagnosis and Treatment (Genetics, Medicine), according to OpenAlex
Funding: Fundação para a Ciência e a Tecnologia; EU funds under projects NEXUS, Instituto de Telecomunicações (UID/50008/2025, UID/50021/2025, UID/50022/2025, LISBOA2030-FEDER-00868200—Projeto Nº 15030, 2023.17447.ICDT)
Citations: not cited yet (Europe PMC); 35 references in the paper

Abstract

Background: Gliomas are the most common malignant brain tumors in adults, characterized by a poor prognosis. Although the current World Health Organization (WHO) classification provides clear guidelines for classifying oligodendroglioma, astrocytoma, and glioblastoma patients, significant heterogeneity persists within each class, limiting the effectiveness of current treatment strategies. With the increasing availability of large-scale multi-omics datasets resulting from advancements in sequencing technologies and online repositories that provide them, such as The Cancer Genome Atlas (TCGA), it is now possible to investigate these tumors at multiple molecular levels. Methods: In this work, we apply integrative multi-omics analysis to explore the interplay between genomic (mutations), epigenomic (DNA methylation), and transcriptomic (mRNA and miRNA) layers. Our approach relies on Multi-Omics Factor Analysis (MOFA), a Bayesian latent factor analysis model designed to capture sources of variation across different omics types. Results: Our results highlight distinct molecular profiles across the three glioma types and identify potential relationships between methylation and genetic expression. In particular, we uncover novel candidate biomarkers associated with survival as well as a transcriptional profile associated with neural system development. Conclusions: These findings may contribute to more personalized therapeutic strategies, potentially improving treatment effectiveness and survival outcomes in this disease.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repository

Its files are read in the Code ↔ Paper reader above, with 4 matches between paragraphs and lines of code.

sysbiomed/MOFA-in-Gliomas

License: none: the authors keep all their rights
State: the link answers, verified on 27 September 2026
Evidence: files inventoried
Commit: 4c8df639fd56d05a48dbcd83f949436403910ba1, 14 April 2026
Languages: R (14)
Size: 69 files, 14 scripts
Software Heritage: not archived
Found in: “Data Availability Statement”
Holds: README, 5 notebooks
Not found: license file, CITATION.cff, environment file, tests, continuous integration, documentation
Tools: tidyverse (4 files), ggplot2 (3 files), DESeq2 (2 files), edgeR (2 files), survival (2 files), circlize (1 file), clusterProfiler (1 file), ComplexHeatmap (1 file), ggpubr (1 file), reshape2 (1 file), reticulate (1 file)
Availability: 1 check, the latest on 27 September 2026: the link answers
  • 27 September 2026: the link answers
15 files
At the source: github.com/sysbiomed/MOFA-in-Gliomas

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

  • 1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
  • 14 scripts, each with its path and the digest of its content;
  • 4 matches between paragraphs of the paper and lines of the code (method lexical-v1);
  • neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

No dataset and no data link were found in the paper.

2.1. Data Availability

The integration in the present study focused on several omics layers, specifically genomics (mutations), as they are key drivers of glioma heterogeneity, along with transcriptomics (mRNA and miRNA) and epigenomics (DNA methylation). The data were obtained from The Cancer Genome Atlas (TCGA) under the project names “TCGA-GBM” and “TCGA-LGG”.

The mutations dataset was downloaded using the RTCGAToolbox (version v2.32.1) [20] package, while the others were obtained using the TCGAbiolinks (version v2.30.4) [21] package. The downloaded data included binary mutation profiles, count-based transcriptomics, and beta values representing the methylation proportion of each probe. These last two datasets were provided in the summarizedExperiment format, which contains not only the expression matrix but also feature metadata, including gene annotations and additional biological information, such as associated chromosomal locations and corresponding gene names. Clinical data, including patient demographics (age, sex) and survival information, were also extracted. Its summary is in Table 1.

The ground-truth glioma labels used were from the study [22], where the TCGA labels were updated according to the most recent WHO guidelines from 2021. The labels assigned were: “Astrocytoma”, “Glioblastoma”, “Oligodendroglioma” or “Unclassified”. Table 2 highlights significant discrepancies between the labels assigned by the WHO and those provided by TCGA for the same patients.

All the code is available at https://github.com/sysbiomed/MOFA-in-Gliomas (accessed on 15 April 2026) to ensure reproducibility and modularity.

Reproduced under the paper's license (CC BY), from the paper cited above.

Data Availability Statement

The R code developed for this analysis is open source and available at https://github.com/sysbiomed/MOFA-in-Gliomas (accessed on 15 April 2026). The original datasets are not included due to their large size, but detailed instructions for downloading them are provided in the repository. The datasets are publicly available from The Cancer Genome Atlas (TCGA) database at https://portal.gdc.cancer.gov. The glioma classification using the WHO-2021 taxonomy guidelines is available at https://github.com/sysbiomed/MONET.

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 27 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 3 authors, 5 keywords, 13 MeSH terms, 2 funders, 34 references.

Cite

This paper

Carvalho, C. G., Carvalho, A. M., & Vinga, S. (2026). Uncovering Latent Structure in Gliomas Using Multi-Omics Factor Analysis. Genes, 17(5), 540. https://doi.org/10.3390/genes17050540

BibTeX

@article{carvalho2026uncovering,
author = {Carvalho, Catarina Gameiro and Carvalho, Alexandra M. and Vinga, Susana},
title = {{Uncovering Latent Structure in Gliomas Using Multi-Omics Factor Analysis}},
journal = {Genes},
year = {2026},
month = may,
volume = {17},
number = {5},
pages = {540},
publisher = {Multidisciplinary Digital Publishing Institute (MDPI)},
issn = {2073-4425},
doi = {10.3390/genes17050540},
url = {https://doi.org/10.3390/genes17050540},
pmid = {42194997},
pmcid = {PMC13205169}
}

RIS

TY - JOUR
AU - Carvalho, Catarina Gameiro
AU - Carvalho, Alexandra M.
AU - Vinga, Susana
TI - Uncovering Latent Structure in Gliomas Using Multi-Omics Factor Analysis
T2 - Genes
J2 - Genes (Basel)
PY - 2026
DA - 2026/05/01
VL - 17
IS - 5
SP - 540
SN - 2073-4425
PB - Multidisciplinary Digital Publishing Institute (MDPI)
DO - 10.3390/genes17050540
UR - https://doi.org/10.3390/genes17050540
LA - en
ER -

CSL-JSON

{
"id": "10.3390/genes17050540",
"type": "article-journal",
"title": "Uncovering Latent Structure in Gliomas Using Multi-Omics Factor Analysis",
"container-title": "Genes",
"author": [
{
"family": "Carvalho",
"given": "Catarina Gameiro"
},
{
"family": "Carvalho",
"given": "Alexandra M."
},
{
"family": "Vinga",
"given": "Susana"
}
],
"container-title-short": "Genes (Basel)",
"volume": "17",
"issue": "5",
"page": "540",
"DOI": "10.3390/genes17050540",
"PMID": "42194997",
"PMCID": "PMC13205169",
"ISSN": "2073-4425",
"publisher": "Multidisciplinary Digital Publishing Institute (MDPI)",
"URL": "https://doi.org/10.3390/genes17050540",
"language": "en",
"issued": {
"date-parts": [
[
2026,
5,
1
]
]
}
}

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1093/bioinformatics/btag592 [code]
Network-based stratification of allele-specific expression reveals patient subgroups in Huntington's disease.
Journal: Bioinformatics (Oxford, England)
In common: survival, edgeR, reticulate, 8 other tools, genetics / omics, other condition
[2] doi:10.1038/s42003-026-10957-8 [code]
Brain defence by the extracellular matrix protein Cochlin.
Journal: Communications biology
In common: survival, edgeR, reticulate, 8 other tools
[3] doi:10.1186/s12967-026-08266-z [code]
Single-cell multi-omic integration analysis prioritizes druggable genes and reveals cell-type-specific causal effects in glioblastomagenesis.
Journal: Journal of translational medicine
In common: edgeR, reticulate, circlize, 6 other tools, genetics / omics, other condition, 2 references
[4] doi:10.1016/j.isci.2026.115657 [code]
Integration of machine learning to develop a disulfidptosis model for predicting glioma prognosis, immunotherapy response, and drug.
Journal: iScience
In common: survival, edgeR, DESeq2, 7 other tools, other condition
[5] doi:10.1038/s41593-026-02367-0 [code]
A reproducible three-dimensional model of human brain tissue to investigate physiological and disease-associated microglia phenotypes.
Journal: Nature neuroscience
In common: edgeR, reticulate, DESeq2, 7 other tools
[6] doi:10.1016/j.xcrm.2026.102766 [code]
A longitudinal single-cell and spatial multiomic atlas of pediatric high-grade glioma.
Journal: Cell reports. Medicine
In common: edgeR, DESeq2, circlize, 6 other tools, genetics / omics, other condition, 1 reference
[7] doi:10.3390/ijms27093997 [code]
Coordinated Multicellular Immune Programs and Drug Targets Revealed by Single-Cell Analysis in Driver-Mutated NSCLC.
Journal: International journal of molecular sciences
In common: survival, DESeq2, circlize, 6 other tools, genetics / omics, other condition
[8] doi:10.1038/s41586-026-10512-9 [code]
Astrocyte glucocorticoid receptor signalling restricts neuronal plasticity.
Journal: Nature
In common: edgeR, DESeq2, circlize, 6 other tools, 1 reference
[9] doi:10.1038/s42003-026-10252-6 [code]
RET signaling as a mediator of estrogen receptor positive breast cancer brain metastasis.
Journal: Communications biology
In common: survival, DESeq2, circlize, 6 other tools, other condition
[10] doi:10.1101/gr.281113.125 [code]
Single-nucleus multiomic profiling of the aging mouse substantia nigra reveals conserved gene alterations linked to Parkinson's disease.
Journal: Genome research
In common: edgeR, reticulate, circlize, 6 other tools, genetics / omics

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.