R/plot.iC10.R

In iC10: A Copy Number and Expression-Based Classifier for Breast Tumours

plot.iC10 <-
function (x, sample.name = 1, newdata = NULL, ...)
{
    par(no.readonly = TRUE)
    oldpar <- par(no.readonly = TRUE)
    if (!is.numeric(sample.name))
        sample.name <- which(names(x$class) == sample.name)
    coliCluster <- c("#FF5500", "#00EE76", "#CD3278", "#00C5CD",
        "#8B0000", "#FFFF40", "#0000CD", "#FFAA00", "#EE82EE",
        "#7D26CD")
    cn.features <- NULL
    exp.features <- NULL
    if (attr(x, "classifier.type") != "Exp") {
        cn.features <- x$map.cn
        cn.features$Type <- "CN"
    }
    if (attr(x, "classifier.type") != "CN") {
        exp.features <- x$map.exp
        exp.features$Type <- "Exp"
    }
    max.grey.square <- nrow(cn.features)
    features <- rbind(cn.features, exp.features)
    if (attr(x, "ref") == "hg18") {
        features$CHROM <- features$chromosome_name_hg18
    }
    if (attr(x, "ref") == "hg19") {
        features$CHROM <- features$chromosome_name_hg19
    }
    Pos <- c(1, which(diff(features$CHROM) != 0), nrow(features))
    text.pos <- Pos + c(diff(Pos)/2, 0)
    text.pos <- text.pos[-length(text.pos)]
    if (!is.null(newdata)) {
        tmp <- rbind(newdata$CN, newdata$Exp)
        for (i in 1:10) {
            if (length(which(x$class == i)) > 0) {
                x$centroids[, i] <- apply(tmp[, which(x$class ==
                  i), drop = F], 1, mean)
            }
        }
    }
    par(mfrow = c(6, 2), mar = c(2.5, 4, 2, 2), oma = c(2, 2,
        2, 2), no.readonly = TRUE)
    for (i in 1:10) {
        plot(x$centroids[, i], type = "h", col = coliCluster[i],
            xlab = "", ylab = paste("iC", i), axes = F, ylim = range(x$centroids))
        axis(2)
        bottom.axis.pos <- seq(from = 1, by = 2, length = length(text.pos))
        top.axis.pos <- seq(from = 2, by = 2, length = length(text.pos))
        axis(1, at = text.pos[bottom.axis.pos], features$CHROM[Pos[-1]][bottom.axis.pos])
        axis(3, at = text.pos[top.axis.pos], features$CHROM[Pos[-1]][top.axis.pos])
        box()
        if (attr(x, "classifier.type") != "Exp") {
            polygon(x = rep(c(0, max.grey.square), c(2, 2)),
                y = c(range(x$centroids[, i]), rev(range(x$centroids[,
                  i]))), density = NA, border = "lightgrey",
                col = "lightgrey", ylab = paste("iC", i))
        }
        points(x$centroids[, i], type = "h", col = coliCluster[i],
            xlab = "")
        abline(v = Pos, lty = 2)
    }
    if (is.null(newdata)) {
        plot(x$fitted[, sample.name], type = "h", col = "black",
            xlab = "", ylab = "Prediction", axes = F)
        axis(2)
        axis(1, at = text.pos[bottom.axis.pos], features$CHROM[Pos[-1]][bottom.axis.pos])
        axis(3, at = text.pos[top.axis.pos], features$CHROM[Pos[-1]][top.axis.pos])
        if (attr(x, "classifier.type") != "Exp") {
            polygon(x = rep(c(0, max.grey.square), c(2, 2)),
                y = c(range(x$fitted[, sample.name]), rev(range(x$fitted[,
                  sample.name]))), density = NA, border = "lightgrey",
                col = "lightgrey")
        }
        points(x$fitted[, sample.name], type = "h", col = "black")
        abline(v = Pos, lty = 2)
        barplot(x$posterior[sample.name, ], col = coliCluster,
            ylim = c(0, 1), ylab = "Prob.classif.")
    }
    else {
        plot(0, 0, type = "n", axes = F, xlab = "", ylab = "")
        barplot(table(x$class), col = coliCluster, ylab = "Frequency")
    }
    par(oldpar, no.readonly = TRUE)
}