R/snaptron_query.R

Defines functions snaptron_query

Documented in snaptron_query

#' Query Snaptron to get data from exon-exon junctions present in Intropolis
#'
#' This function uses the Snaptron API to query specific exon-exon junctions
#' that are available via Intropolis as described in the vignette.
#'
#' @param junctions A [GRanges-class][GenomicRanges::GRanges-class] object with the
#' exon-exon junctions of interest. The chromosome names should be in UCSC
#' format, such as 'chr1'. The strand information is ignored in the query.
#' @param version Either `srav1`, `srav2`, `gtex` or
#' `tcga`. SRA Version 1 of Intropolis has the
#' exon-exon junctions from about 20 thousand RNA-seq samples in hg19
#' coordinates. SRA Version 2 has the data from about 50 thousand RNA-seq
#' samples aligned to hg38. GTEx has about 30 million junctions from about 10
#' thousand samples from the GTEx consortium on hg38 coordinates. Finally,
#' TCGA has about 36 million junctions from about 11 thousand samples
#' from the TCGA consortium on hg38 coordinates.
#' @param verbose If `TRUE` status updates will be printed.
#' @param async Defaults to `TRUE` but in some situations it might be
#' preferrable to set it to `FALSE`. This argument gets passed to
#' [getURL][RCurl::getURL]. Check <https://github.com/ChristopherWilks/snaptron/issues/11>
#' for more details.
#'
#' @return A [GRanges-class][GenomicRanges::GRanges-class] object with the results from
#' the Snaptron query. For information on the different columns please see
#' <http://snaptron.cs.jhu.edu>.
#'
#' @references Please cite <http://snaptron.cs.jhu.edu>
#' if you use this function as Snaptron is a separate project from recount.
#' Thank you!
#'
#' @author Leonardo Collado-Torres
#' @export
#'
#' @import GenomicRanges
#' @import RCurl
#'
#' @examples
#'
#' library("GenomicRanges")
#' ## Define some exon-exon junctions (hg19 coordinates)
#' junctions <- GRanges(seqnames = "chr2", IRanges(
#'     start = c(28971710:28971712, 29555081:29555083, 29754982:29754984),
#'     end = c(29462417:29462419, 29923338:29923340, 29917714:29917716)
#' ))
#'
#' ## Check against Snaptron SRA version 1 (hg19 coordinates)
#' snaptron_query(junctions)
#' \dontrun{
#' ## Check another set of junctions against SRA version 2 (more data, hg38
#' ## coordinates)
#' junctions_v2 <- GRanges(seqnames = "chr2", IRanges(
#'     start = 29532116:29532118, end = 29694848:29694850
#' ))
#' snaptron_query(junctions_v2, version = "srav2")
#'
#' ## Check these junctions in GTEx and TCGA data
#' snaptron_query(junctions_v2, version = "gtex")
#' snaptron_query(junctions_v2, version = "tcga")
#' }
#'
snaptron_query <- function(junctions, version = "srav1", verbose = TRUE, async = TRUE) {
    ## Check input
    stopifnot(is(junctions, "GRanges"))
    stopifnot(all(grepl("chr", seqlevels(junctions))))
    version <- tolower(version)
    stopifnot(version %in% c("srav1", "srav2", "gtex", "tcga"))

    ## Build query URLs
    urls <- paste0(
        "http://snaptron.cs.jhu.edu/", version,
        "/snaptron?regions=", seqnames(junctions), ":", start(junctions), "-",
        end(junctions), "&exact=1&header=0"
    )

    ## Get results
    if (verbose) message(paste(Sys.time(), "querying Snaptron"))
    query_res <- getURL(urls, async = async)

    ## Split by line
    if (verbose) message(paste(Sys.time(), "processing results"))
    query_split <- strsplit(query_res, "\n")
    names(query_split) <- NULL

    ## Are there any valid ones?
    valid <- which(elementNROWS(query_split) > 0)
    if (length(valid) == 0) {
        message(paste(
            Sys.time(),
            "found no exon-exon junctions in Intropolis version", version,
            "matching your query: this version uses",
            ifelse(version == "srav1", "hg19", "hg38"), "coordinates."
        ))
        return(NULL)
    }

    ## Extract information
    if (verbose) message(paste(Sys.time(), "extracting information"))
    info <- lapply(query_split[valid], function(jxs) {
        matrix(strsplit(jxs, "\t")[[1]], ncol = 18)
    })

    ## Build result
    res <- do.call(rbind, info)
    colnames(res) <- c(
        "type", "snaptron_id", "chromosome", "start", "end",
        "length", "strand", "annotated", "left_motif", "right_motif",
        "left_annotated", "right_annotated", "samples",
        "samples_count", "coverage_sum", "coverage_avg", "coverage_median",
        "source_dataset_id"
    )

    ## Helper function for some special variables
    to_chr_list <- function(x) {
        r <- strsplit(x, ",")
        i <- which(sapply(r, function(y) {
            y[[1]] == "0"
        }))
        if (length(i) > 0) r[i] <- NA
        return(CharacterList(r))
    }

    ## Build into a GRanges object
    result <- GRanges(
        seqnames = res[, "chromosome"],
        IRanges(as.numeric(res[, "start"]), as.numeric(res[, "end"])),
        strand = res[, "strand"]
    )

    ## Add other variables
    result$type <- as.factor(res[, "type"])
    result$snaptron_id <- as.integer(res[, "snaptron_id"])
    result$annotated <- to_chr_list(res[, "annotated"])
    result$left_motif <- res[, "left_motif"]
    result$right_motif <- res[, "right_motif"]
    result$left_annotated <- to_chr_list(res[, "left_annotated"])
    result$right_annotated <- to_chr_list(res[, "right_annotated"])

    ## Remove leading comma
    res[, "samples"] <- gsub("^,", "", res[, "samples"])
    ## Split sample ids from the read coverage by sample
    sample_info <- IntegerList(strsplit(res[, "samples"], ":|,"))
    sample_idx <- LogicalList(lapply(elementNROWS(sample_info), function(y) {
        rep(c(TRUE, FALSE), y / 2)
    }))
    result$samples <- sample_info[sample_idx]
    result$read_coverage_by_sample <- sample_info[!sample_idx]

    ## Continue with other variables
    result$samples_count <- as.integer(res[, "samples_count"])
    result$coverage_sum <- as.integer(res[, "coverage_sum"])
    result$coverage_avg <- as.numeric(res[, "coverage_avg"])
    result$coverage_median <- as.numeric(res[, "coverage_median"])
    result$source_dataset_id <- as.integer(res[, "source_dataset_id"])

    ## Finish
    return(result)
}

Try the recount package in your browser

Any scripts or data that you put into this service are public.

recount documentation built on Dec. 20, 2020, 2:01 a.m.