R/standardize.R

In RajLabMSSM/echodata: echoverse module: Fine-mapping data access and formatting

#' Standardize the locus subset
#'
#' After querying a subset of the full summary statistics,
#' this function converts it into a standardized format
#' that the rest of \emph{echolocatoR} can work with.
#'
#' @param query data.frame of summary stat.
#' @param subset_path Path where the \code{query} should be 
#' saved after standardization.
#' @param locus Locus name.
#' @param colmap Column mapping object created using 
#' \link[echodata]{construct_colmap}. 
#' @param return_dt Return data.table or path to saved data.table.
#' @param nThread Number of threads to parallelise saving across.
#' @param verbose Print messages.
#' @inheritParams get_sample_size
#' @inheritParams MungeSumstats::compute_nsize
#'
#' @export 
#' @importFrom dplyr rename mutate arrange mutate_at 
#' @importFrom data.table fread data.table fwrite
#' @examples
#' query <- echodata::BST1
#' #### Screw up data
#' query$rsid <- query$SNP
#' query <- query[,-c("MAF","SNP")]
#' query[c(10,30,55),"Freq"] <- 0
#' query[c(12,22),"Freq"] <- NA 
#' subset_path <- file.path(tempdir(),"BST1.tsv")
#' colmap = construct_colmap(SNP="rsid")
#' query2 <- echodata::standardize(query = query,
#'                                 subset_path = subset_path,
#'                                 locus = "BST1",
#'                                 colmap = colmap)
standardize <- function(query,
                        subset_path = NULL,
                        locus = NULL,
                        colmap = construct_colmap(),
                        compute_n = "ldsc",
                        return_dt = TRUE,
                        nThread = 1, 
                        verbose = TRUE){ 
   
    query <- data.table::data.table(query)
    messager("Standardizing summary statistics subset.",v=verbose) 
    # ------ Required columns ------ #
    #### Rename main cols ####
    query <- standardize_main_colnames(query=query,
                                       colmap=colmap, 
                                       verbose=verbose)
    #### Impute StdErr ####
    query <- standardize_stderr(query=query,
                                colmap=colmap,
                                verbose=verbose)
    #### Gene col ####
    query <- standardize_gene_col(query=query, 
                                  colmap=colmap,
                                  locus=locus,
                                  verbose=verbose) 
    # ------ Optional columns ------ # 
    #### Add ref/alt alleles if available ####
    query <- standardize_a1_a2(query=query,
                               colmap=colmap,
                               verbose=verbose) 
    ### MAF ####
    query <- standardize_maf(query=query, 
                             colmap=colmap,
                             verbose=verbose)
    #### N_cases/N_controls ###
    query <- standardize_ncas_ncon(query=query,
                                   colmap=colmap,
                                   verbose=verbose)
    #### proportion_cases ####
    query <- standardize_proportion_cases(query=query, 
                                         colmap=colmap, 
                                         verbose=verbose)
    #### sample size ####  
    query <- get_sample_size(dat=query,
                             compute_n=compute_n, 
                             force_new=FALSE,
                             verbose=verbose)
    #### Impute t-stat #### 
    ## Assign renamed values  
    colmap$Effect <- "Effect"; colmap$StdErr <- "StdErr"
    ## Impute 
    query <- calculate_tstat(dat=query,
                             colmap=colmap,
                             verbose=verbose)
    #### P ####
    query <- standardize_p(query=query,
                           verbose=verbose) 
    #### Remove duplicate columns ####
    query <- data.frame(query)[,!duplicated(colnames(query))]
    #### leadSNP ####
    query <- assign_lead_snp(dat = query, 
                             verbose = verbose)  
    ## Only convert to numeric AFTER removing NAs
    ## (otherwise as.numeric will turn them into 0s)
    query <- standardize_coltypes(query=query, 
                                  verbose=verbose) 
    ## Get just one SNP per location (just pick the first one)
    query <- standardize_1snp_per_coord(query=query,
                                        verbose=verbose)
    # Trim whitespaces
    ## Extra whitespace causes problems when you try
    ## to make space-delimited files.
    query <- standardize_trimws(query=query, 
                                verbose=verbose)
    #### Report #### 
    messager("+ Standardized query:",
             formatC(dim(query)[1], big.mark = ","), "SNPs",
             "x",
             formatC(dim(query)[2],big.mark = ","),
             "columns.",
             v=verbose)
    #### Save ####
    if(!is.null(subset_path)){
        messager("++ Saving standardized query ==>",subset_path,
                 v=verbose)
        dir.create(dirname(subset_path), 
                   showWarnings = FALSE, recursive = TRUE)
        data.table::fwrite(query,
                           subset_path,
                           sep = "\t",
                           nThread = nThread)
    } 
    #### Return ####
    if(isFALSE(return_dt) && !is.null(subset_path)){
        return(subset_path) 
    } else { 
        return(data.table::data.table(query))
    } 
}