biopipen 0.33.1__py3-none-any.whl → 0.34.1__py3-none-any.whl

biopipen/utils/gsea.R

@@ -1,329 +0,0 @@
-library(ggplot2)
-library(dplyr)
-library(tibble)
-
-if (!exists("slugify")) {
-    slugify <- function(x, non_alphanum_replace="-", collapse_replace=TRUE, tolower=FALSE) {
-        subs <- list(
-            "š"="s", "œ"="oe", "ž"="z", "ß"="ss", "þ"="y", "à"="a", "á"="a", "â"="a",
-            "ã"="a", "ä"="a", "å"="a", "æ"="ae", "ç"="c", "è"="e", "é"="e", "ê"="e",
-            "ë"="e", "ì"="i", "í"="i", "î"="i", "ï"="i", "ð"="d", "ñ"="n", "ò"="o",
-            "ó"="o", "ô"="o", "õ"="o", "ö"="o", "ø"="oe", "ù"="u", "ú"="u", "û"="u",
-            "ü"="u", "ý"="y", "ÿ"="y", "ğ"="g", "ı"="i", "ij"="ij", "ľ"="l", "ň"="n",
-            "ř"="r", "ş"="s", "ť"="t", "ų"="u", "ů"="u", "ý"="y", "ź"="z", "ż"="z",
-            "ſ"="s", "α"="a", "β"="b", "γ"="g", "δ"="d", "ε"="e", "ζ"="z", "η"="h",
-            "θ"="th", "ι"="i", "κ"="k", "λ"="l", "μ"="m", "ν"="n", "ξ"="x", "ο"="o",
-            "π"="p", "ρ"="r", "σ"="s", "τ"="t", "υ"="u", "φ"="ph", "χ"="ch", "ψ"="ps",
-            "ω"="o", "ά"="a", "έ"="e", "ή"="h", "ί"="i", "ό"="o", "ύ"="u", "ώ"="o",
-            "ϐ"="b", "ϑ"="th", "ϒ"="y", "ϕ"="ph", "ϖ"="p", "Ϛ"="st", "ϛ"="st", "Ϝ"="f",
-            "ϝ"="f", "Ϟ"="k", "ϟ"="k", "Ϡ"="k", "ϡ"="k", "ϰ"="k", "ϱ"="r", "ϲ"="s",
-            "ϳ"="j", "ϴ"="th", "ϵ"="e", "϶"="p"
-        )
-        # replace latin and greek characters to the closest english character
-        for (k in names(subs)) {
-            x <- gsub(k, subs[[k]], x)
-        }
-        x <- gsub("[^[:alnum:]_]", non_alphanum_replace, x)
-        if(collapse_replace) x <- gsub(
-            paste0(gsub("([][{}()+*^$|\\\\?.])", "\\\\\\1", non_alphanum_replace), "+"),
-            non_alphanum_replace,
-            x
-        )
-        if(tolower) x <- tolower(x)
-        x
-    }
-}
-
-#' Download the GMT file and save it to cachedir
-#' Return the path to the GMT file
-#' We also check if the second column is shorter than the first column.
-#' If so, we switch the first and second columns.
-#' In case some providers provide the GMT file with the first and second columns switched.
-#' We also replace the "/" in the gene names with "-" if any. This is because the "/" is
-#' not allowed in a path, but GSEA uses the gene names as the file name.
-#'
-#' @param gmturl The URL or path of the GMT file
-#' @param cachedir The directory to save the GMT file
-#' @return The path to the GMT file
-localizeGmtfile <- function(gmturl, cachedir = tempdir()) {
-    # Download the GMT file and save it to cachedir
-    # Return the path to the GMT file
-    in_gmtfile <- out_gmtfile <- file.path(cachedir, basename(gmturl))
-    if (startsWith(gmturl, "http") || startsWith(gmturl, "ftp")) {
-        download.file(gmturl, in_gmtfile)
-        remote <- TRUE
-    } else {
-        in_gmtfile <- gmturl
-        remote <- FALSE
-    }
-
-    items <- readLines(in_gmtfile)
-    items <- items[!grepl("^#", items) & nchar(items) > 0]
-    items <- lapply(strsplit(items, "\t"), function(x) c(x[1:2], paste0(x[3:length(x)], collapse = "\t")))
-    items <- as.data.frame(t(as.data.frame(items)))
-    rownames(items) <- NULL
-    colnames(items) <- c("V1", "V2", "V3")
-
-    if (ncol(items) < 3) {
-        stop(paste0("Invalid GMT file: ", gmturl))
-    }
-    if (nrow(items) == 0) {
-        stop(paste0("Empty GMT file: ", gmturl))
-    }
-
-    # Check if the second column is shorter than the first column
-    nchars1 <- sum(nchar(items$V1))
-    nchars2 <- sum(nchar(items$V2))
-    prefix <- gsub("[0-9]+$", "", items$V2[1])
-
-    if (is.character(items$V2) &&     # Only when V2 is character, as pathway names
-        nchars2 < nchars1 &&          # Only when V2 is shorter than V1
-        all(nchar(items$V2) > 0) &&   # Only when V2 is not empty
-        !all(grepl("^[0-9]+$", items$V2)) && # Only when V2 is not all numbers
-        (nchar(prefix) == 0 || !all(startsWith(items$V2, prefix))) # Only when they are not like hsa00001, hsa00002, etc.
-    ) {
-        warning(paste0(
-            "The second column is shorter, switching the first and second columns in ",
-            "GMT file ", gmturl
-        ))
-        items <- items[, c(2, 1, 3:ncol(items))]
-        switched <- TRUE
-    } else {
-        switched <- FALSE
-    }
-
-    if (any(grepl("/", items$V1))) {
-        items$V1 <- gsub("/", "-", items$V1)
-        replaced <- TRUE
-    } else {
-        replaced <- FALSE
-    }
-
-    if (remote || switched || replaced) {
-        write.table(
-            items,
-            out_gmtfile,
-            row.names = FALSE,
-            col.names = FALSE,
-            sep = "\t",
-            quote = FALSE
-        )
-    } else {
-        out_gmtfile <- in_gmtfile
-    }
-
-    return(out_gmtfile)
-}
-
-
-prerank <- function(
-    exprdata,
-    pos,
-    neg,
-    classes, # must be in the order of colnames(exprdata)
-    method = "signal_to_noise"
-) {
-    library(matrixStats)
-    set.seed(8525)
-    # See: https://gseapy.readthedocs.io/en/latest/_modules/gseapy/algorithm.html#ranking_metric
-    expr_pos_mean = rowMeans(exprdata[, classes == pos, drop=F], na.rm=TRUE)
-    expr_neg_mean = rowMeans(exprdata[, classes == neg, drop=F], na.rm=TRUE)
-    expr_pos_std = rowSds(as.matrix(exprdata[, classes == pos, drop=F]), na.rm=TRUE, useNames = T)
-    expr_neg_std = rowSds(as.matrix(exprdata[, classes == neg, drop=F]), na.rm=TRUE, useNames = T)
-    rands = rnorm(length(expr_neg_std)) * 1e-6
-
-    if (method %in% c("s2n", "signal_to_noise")) {
-        out = (expr_pos_mean - expr_neg_mean) / (expr_pos_std + expr_neg_std + rands)
-    } else if (method %in% c("abs_s2n", "abs_signal_to_noise")) {
-        out = abs((expr_pos_mean - expr_neg_mean) / (expr_pos_std + expr_neg_std + rands))
-    } else if (method == "t_test") {
-        # ser = (df_mean[pos] - df_mean[neg])/ np.sqrt(df_std[pos]**2/len(df_std)+df_std[neg]**2/len(df_std) )
-        out = (expr_pos_mean - expr_neg_mean) / sqrt(
-            expr_pos_std ^ 2 / length(expr_pos_std) +
-            expr_neg_std ^ 2 / length(expr_neg_std)
-        )
-    } else if (method == "ratio_of_classes") {
-        out = expr_pos_mean / expr_neg_mean
-    } else if (method == "diff_of_classes") {
-        out = expr_pos_mean - expr_neg_mean
-    } else if (method == "log2_ratio_of_classes") {
-        out = log2(expr_pos_mean) - log2(expr_neg_mean)
-    } else {
-        stop(paste("Unknown method:", method))
-    }
-    # todo: log2fc * -log10(p)
-    # see https://github.com/crazyhottommy/RNA-seq-analysis/blob/master/GSEA_explained.md#2-using-a-pre-ranked-gene-list
-    out = as.data.frame(out) %>% rownames_to_column("Gene") %>% arrange(.[[2]])
-    colnames(out)[2] = paste(pos, "vs", neg, sep="_")
-    return(out)
-}
-
-runEnrichr = function(
-    genes,
-    dbs,
-    outdir,
-    showTerms = 20,
-    numChar =40,
-    orderBy = "P.value"
-) {
-    library(enrichR)
-    setEnrichrSite("Enrichr") # Human genes
-
-    enriched = enrichr(genes, dbs)
-
-    for (db in dbs) {
-        enr = enriched[[db]] %>% select(-c(Old.P.value, Old.Adjusted.P.value))
-        outtable = file.path(outdir, paste0("Enrichr_", db, ".txt"))
-        outfig = file.path(outdir, paste0("Enrichr_", db, ".png"))
-        write.table(enr, outtable, row.names=T, col.names=F, sep="\t", quote=F)
-
-        if (nrow(enr) == 0) {
-            print(paste0("No enriched terms for ", db))
-            next
-        }
-
-        png(outfig, res=100, height=1000, width=1400)
-        print(
-            plotEnrich(
-                enriched[[db]],
-                showTerms=showTerms,
-                numChar=numChar,
-                orderBy=orderBy
-            )
-        )
-        dev.off()
-    }
-
-}
-
-runFGSEA = function(
-    ranks,
-    gmtfile,
-    top,
-    outdir,
-    envs = list(),
-    plot = TRUE  # only generate fgsea.txt?
-) {
-    library(data.table)
-    library(fgsea)
-    set.seed(8525)
-
-    if (is.data.frame(ranks)) {
-        ranks = setNames(ranks[[2]], ranks[[1]])
-    } else if (is.list(ranks)) {
-        ranks = unlist(ranks)
-    }
-
-    gmtfile = localizeGmtfile(gmtfile)
-    envs$pathways = gmtPathways(gmtfile)
-    envs$stats = ranks
-    gsea_res = do.call(fgsea::fgsea, envs)
-    gsea_res = gsea_res[order(pval), ]
-
-    write.table(
-        gsea_res %>%
-            mutate(leadingEdge = sapply(leadingEdge, function(x) paste(x, collapse=",")),
-                   slug = sapply(pathway, slugify)),
-        file = file.path(outdir, "fgsea.txt"),
-        row.names = FALSE,
-        col.names = TRUE,
-        sep = "\t",
-        quote = FALSE
-    )
-
-    if (!plot) {return (NULL)}
-
-    if (top > 1) {
-        topPathways = head(gsea_res, n=top)[, "pathway"]
-    } else {
-        topPathways = gsea_res[padj < top][, "pathway"]
-    }
-    topPathways = unlist(topPathways)
-
-    p <- plotGseaTable(
-        envs$pathways[topPathways],
-        ranks,
-        gsea_res,
-        gseaParam = if (!is.null(envs$gseaParam)) envs$gseaParam else 1
-    )
-
-    tablefig = file.path(outdir, "gsea_table.png")
-    png(tablefig, res=100, width=1000, height=200 + 40 * length(topPathways))
-    print(p)
-    dev.off()
-
-    tablefig_pdf = file.path(outdir, "gsea_table.pdf")
-    pdf(tablefig_pdf, width=10, height=2 + 0.4 * length(topPathways))
-    print(p)
-    dev.off()
-
-    for (pathway in topPathways) {
-        enrfig = file.path(outdir, paste0("fgsea_", slugify(pathway), ".png"))
-        p <- plotEnrichment(
-            envs$pathways[[pathway]],
-            ranks,
-            gseaParam = if (!is.null(envs$gseaParam)) envs$gseaParam else 1
-        ) + labs(title = pathway)
-
-        png(enrfig, res=100, width=1000, height=800)
-        print(p)
-        dev.off()
-
-        enrfig_pdf = gsub(".png$", ".pdf", enrfig)
-        pdf(enrfig_pdf, width=10, height=8)
-        print(p)
-        dev.off()
-    }
-}
-
-runGSEA = function(
-    indata, # expression data
-    classes, # sample classes
-    gmtfile, # the GMT file
-    outdir,
-    envs = list() # other arguments for GSEA()
-) {
-    library(GSEA)
-    # reproducibility
-    if (is.null(envs$random.seed)) {
-        envs$random.seed <- 8525
-    }
-
-    # prepare gct file
-    gctfile = file.path(outdir, "gsea.gct")
-    con = file(gctfile, open='w')
-    write("#1.2", con)
-    write(paste(dim(indata), collapse = "\t"), con)
-    close(con)
-    indata = indata %>%
-        as.data.frame() %>%
-        mutate(Description = "na") %>%
-        rownames_to_column("NAME") %>%
-        select(NAME, Description, everything())
-
-    write.table(
-        indata,
-        gctfile,
-        row.names = F,
-        col.names = T,
-        sep="\t",
-        quote=F,
-        append = T
-    )
-
-    # prepare cls file
-    clsfile = file.path(outdir, "gsea.cls")
-    uniclasses = unique(classes)
-    con = file(clsfile, open='w')
-    write(paste(length(classes), length(uniclasses), '1'), con)
-    write(paste('#', paste(uniclasses, collapse=" ")), con)
-    write(paste(classes, collapse=" "), con)
-    close(con)
-
-    envs$input.ds = gctfile
-    envs$input.cls = clsfile
-    envs$gs.db = localizeGmtfile(gmtfile)
-    envs$output.directory = outdir
-
-    do.call(GSEA, envs)
-}

biopipen/utils/io.R

@@ -1,20 +0,0 @@
-read.table.opts = function(file, opts) {
-    rncol = NULL
-    if (!is.null(opts$row.names) && opts$row.names < 0) {
-        rncol = -opts$row.names
-        opts$row.names = NULL
-        opts = c(opts, list(row.names=NULL))
-    }
-    if (endsWith(file, ".gz")) {
-        opts$file = gzfile(file)
-    } else {
-        opts$file = file
-    }
-    out = do.call(read.table, opts)
-    if (!is.null(rncol)) {
-        rnames = make.unique(out[, rncol])
-        out = out[, -rncol, drop=F]
-        rownames(out) = rnames
-    }
-    return (out)
-}