comparison mutational_patterns.R @ 1:c73aa7cc4c4b draft

"planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit 6ca5597637439c87b61af2dbd6c38089b29eca37"

0:d585f369bfad

    # Assign signature names
    colnames(nmf_res$signatures) <- paste0("NewSig_", 1:opt$newsignum)
    rownames(nmf_res$contribution) <- paste0("NewSig_", 1:opt$newsignum)
    # Plot the 96-profile of the signatures:
    p5 <- plot_96_profile(nmf_res$signatures, condensed = TRUE)
    new_sig_matrix <- reshape2::dcast(p5$data, substitution + context ~ variable, value.var = "value")
    new_sig_matrix <- format(new_sig_matrix, scientific = TRUE)
    write.table(new_sig_matrix, file = opt$sigmatrix, quote = FALSE, row.names = FALSE, sep = "\t")
    grid.arrange(p5)
    # Visualize the contribution of the signatures in a barplot
    pc1 <- plot_contribution(nmf_res$contribution, nmf_res$signature, mode = "relative", coord_flip = TRUE)

        row.names(cancer_signatures) <- cancer_signatures$Somatic.Mutation.Type
        cancer_signatures <- as.matrix(cancer_signatures[, 4:33])
        colnames(cancer_signatures) <- gsub("Signature.", "", colnames(cancer_signatures)) # shorten signature labels
        cosmic_tag <- "Signatures (Cosmic v2, March 2015)"
        cosmic_colors <- col_vector[1:30]
        } else {
        sp_url <- "https://raw.githubusercontent.com/ARTbio/startbio/master/sigProfiler_SBS_signatures_2019_05_22.tsv"
        cancer_signatures <- read.table(sp_url, sep = "\t", header = TRUE)
        new_order <- match(row.names(pseudo_mut_mat), cancer_signatures$Somatic.Mutation.Type)
        cancer_signatures <- cancer_signatures[as.vector(new_order), ]
        row.names(cancer_signatures) <- cancer_signatures$Somatic.Mutation.Type
        cancer_signatures <- as.matrix(cancer_signatures[, 4:70])
        colnames(cancer_signatures) <- gsub("SBS", "", colnames(cancer_signatures)) # shorten signature labels
        cosmic_tag <- "Signatures (Cosmic v3, May 2019)"
        cosmic_colors <- col_vector[1:67]
        }

    # Plot mutational profiles of the COSMIC signatures
    if (opt$cosmic_version == "v2") {
        p6 <- plot_96_profile(cancer_signatures, condensed = TRUE, ymax = 0.3)

    # Find optimal contribution of COSMIC signatures to reconstruct 96 mutational profiles
    fit_res <- fit_to_signatures(pseudo_mut_mat, cancer_signatures)

    # Plot contribution barplots
        pc3 <- plot_contribution(fit_res$contribution, cancer_signatures, coord_flip = T, mode = "absolute")
        pc4 <- plot_contribution(fit_res$contribution, cancer_signatures, coord_flip = T, mode = "relative")
        pc3_data <- pc3$data
        pc3 <- ggplot(pc3_data, aes(x = Sample, y = Contribution, fill = as.factor(Signature))) +
               geom_bar(stat = "identity", position = "stack") +
               coord_flip() +
               scale_fill_manual(name = "Cosmic\nSignatures", values = cosmic_colors) +

               scale_y_continuous(labels = scales::percent_format(accuracy = 1)) +
               labs(x = "Samples", y = "Relative contribution") + theme_bw() +
               theme(panel.grid.minor.x = element_blank(), panel.grid.major.x = element_blank(), legend.position = "right",
                     text = element_text(size = 8),
                     axis.text.x = element_text(angle = 90, hjust = 1))
    #####
    # ggplot2 alternative
    if (!is.na(opt$levels)[1]) {  # if there are levels to display in graphs
        pc3_data <- pc3$data
        pc3_data <- merge(pc3_data, metadata_table[, c(1, 3)], by.x = "Sample", by.y = "element_identifier")

             worklist[[i]] <- worklist[[i]][1:opt$signum, ]
             worklist[[i]] <- worklist[[i]][, -length(as.data.frame(worklist[[i]]))]
        }
        worklist <- as.data.frame(melt(worklist))
        worklist[, 2] <- paste0(worklist[, 4], " - ", worklist[, 2])
        head(worklist)
    }

    colnames(worklist) <- c("signature", "sample", "value", "level")
    worklist <- as.data.frame(worklist %>% group_by(sample) %>% mutate(value = value / sum(value) * 100))
    worklist$pos <- cumsum(worklist$value) - worklist$value / 2

              geom_bar(width = 1, stat = "identity") +
              geom_text(aes(label = label), position = position_stack(vjust = 0.5), color = "black", size = 3) +
              coord_polar("y", start = 0) + facet_wrap(.~sample) +
              labs(x = "", y = "Samples", fill = cosmic_tag) +
              scale_fill_manual(name = paste0(opt$signum, " most contributing\nsignatures\n(in each label/tissue)"),
                                values = cosmic_colors[as.numeric(levels(factor(worklist$label)))]) +
              theme(axis.text = element_blank(),
                    axis.ticks = element_blank(),
                    panel.grid  = element_blank())
    grid.arrange(p7)

1:c73aa7cc4c4b

    # Assign signature names
    colnames(nmf_res$signatures) <- paste0("NewSig_", 1:opt$newsignum)
    rownames(nmf_res$contribution) <- paste0("NewSig_", 1:opt$newsignum)
    # Plot the 96-profile of the signatures:
    p5 <- plot_96_profile(nmf_res$signatures, condensed = TRUE)
    new_sig_matrix <- reshape2::dcast(p5$data, substitution + context ~ sample, value.var = "freq")
    new_sig_matrix <- format(new_sig_matrix, scientific = TRUE)
    write.table(new_sig_matrix, file = opt$sigmatrix, quote = FALSE, row.names = FALSE, sep = "\t")
    grid.arrange(p5)
    # Visualize the contribution of the signatures in a barplot
    pc1 <- plot_contribution(nmf_res$contribution, nmf_res$signature, mode = "relative", coord_flip = TRUE)

        row.names(cancer_signatures) <- cancer_signatures$Somatic.Mutation.Type
        cancer_signatures <- as.matrix(cancer_signatures[, 4:33])
        colnames(cancer_signatures) <- gsub("Signature.", "", colnames(cancer_signatures)) # shorten signature labels
        cosmic_tag <- "Signatures (Cosmic v2, March 2015)"
        cosmic_colors <- col_vector[1:30]
        names(cosmic_colors) <- colnames(cancer_signatures)
        } else {
        sp_url <- "https://raw.githubusercontent.com/ARTbio/startbio/master/sigProfiler_SBS_signatures_2019_05_22.tsv"
        cancer_signatures <- read.table(sp_url, sep = "\t", header = TRUE)
        new_order <- match(row.names(pseudo_mut_mat), cancer_signatures$Somatic.Mutation.Type)
        cancer_signatures <- cancer_signatures[as.vector(new_order), ]
        row.names(cancer_signatures) <- cancer_signatures$Somatic.Mutation.Type
        cancer_signatures <- as.matrix(cancer_signatures[, 4:70])
        colnames(cancer_signatures) <- gsub("SBS", "", colnames(cancer_signatures)) # shorten signature labels
        cosmic_tag <- "Signatures (Cosmic v3, May 2019)"
        cosmic_colors <- col_vector[1:67]
        names(cosmic_colors) <- colnames(cancer_signatures)
        }

    # Plot mutational profiles of the COSMIC signatures
    if (opt$cosmic_version == "v2") {
        p6 <- plot_96_profile(cancer_signatures, condensed = TRUE, ymax = 0.3)

    # Find optimal contribution of COSMIC signatures to reconstruct 96 mutational profiles
    fit_res <- fit_to_signatures(pseudo_mut_mat, cancer_signatures)

    # Plot contribution barplots
    pc3 <- plot_contribution(fit_res$contribution, cancer_signatures, coord_flip = T, mode = "absolute")
    pc4 <- plot_contribution(fit_res$contribution, cancer_signatures, coord_flip = T, mode = "relative")
    if (is.na(opt$levels)[1]) {  # if there are NO levels to display in graphs
        pc3_data <- pc3$data
        pc3 <- ggplot(pc3_data, aes(x = Sample, y = Contribution, fill = as.factor(Signature))) +
               geom_bar(stat = "identity", position = "stack") +
               coord_flip() +
               scale_fill_manual(name = "Cosmic\nSignatures", values = cosmic_colors) +

               scale_y_continuous(labels = scales::percent_format(accuracy = 1)) +
               labs(x = "Samples", y = "Relative contribution") + theme_bw() +
               theme(panel.grid.minor.x = element_blank(), panel.grid.major.x = element_blank(), legend.position = "right",
                     text = element_text(size = 8),
                     axis.text.x = element_text(angle = 90, hjust = 1))
    }
    #####
    # ggplot2 alternative
    if (!is.na(opt$levels)[1]) {  # if there are levels to display in graphs
        pc3_data <- pc3$data
        pc3_data <- merge(pc3_data, metadata_table[, c(1, 3)], by.x = "Sample", by.y = "element_identifier")

             worklist[[i]] <- worklist[[i]][1:opt$signum, ]
             worklist[[i]] <- worklist[[i]][, -length(as.data.frame(worklist[[i]]))]
        }
        worklist <- as.data.frame(melt(worklist))
        worklist[, 2] <- paste0(worklist[, 4], " - ", worklist[, 2])
    }

    colnames(worklist) <- c("signature", "sample", "value", "level")
    worklist <- as.data.frame(worklist %>% group_by(sample) %>% mutate(value = value / sum(value) * 100))
    worklist$pos <- cumsum(worklist$value) - worklist$value / 2

              geom_bar(width = 1, stat = "identity") +
              geom_text(aes(label = label), position = position_stack(vjust = 0.5), color = "black", size = 3) +
              coord_polar("y", start = 0) + facet_wrap(.~sample) +
              labs(x = "", y = "Samples", fill = cosmic_tag) +
              scale_fill_manual(name = paste0(opt$signum, " most contributing\nsignatures\n(in each label/tissue)"),
                                values = cosmic_colors[levels(worklist$signature)]) +
              theme(axis.text = element_blank(),
                    axis.ticks = element_blank(),
                    panel.grid  = element_blank())
    grid.arrange(p7)