Enhancement and refactoring about rain

[mcanouil]

Note: PLINK is used to convert from VCF to BED/BIM/FAM for flashpcaR to use directly.

[mcanouil]

Notes:

• VCFtools (and awk) to be replaced by BCFtools call:

  • rain/R/format_vcf.R

    Lines 41 to 48 in dbfa02c

    	bin_path[["vcftools"]],
    	"--gzvcf", ref1kg_vcfs,
    	"--maf", ref1kg_maf,
    	"--recode",
    	"--stdout",
    	"|", bin_path[["bgzip"]], "-c >", output_ref,
    	"&&",
    	bin_path[["tabix"]], "-p vcf", output_ref

  • rain/R/format_vcf.R

    Lines 72 to 79 in dbfa02c

    	paste(bin_path[["vcftools"]],
    	"--gzvcf", input_vcfs,
    	"--get-INFO", quality_tag,
    	"--out", gsub("filtered_", "excluded_", output_study_temp),
    	"&&",
    	'awk \'{if($5<', quality_threshold, ') print $1"\t"$2}\'',
    	paste0(gsub("filtered_", "excluded_", output_study_temp), ".INFO"),
    	">", paste0(gsub("filtered_", "excluded_", output_study_temp), ".exclude"),

  • rain/R/format_vcf.R

    Lines 83 to 95 in dbfa02c

    	bin_path[["vcftools"]],
    	"--gzvcf", input_vcfs,
    	if (!is.null(quality_tag)) {
    	paste0(
    	"--exclude-positions ", gsub("filtered_", "excluded_", output_study_temp), ".exclude"
    	)
    	},
    	"--remove-indels",
    	"--remove-filtered-all",
    	"--max-missing-count 1",
    	"--recode",
    	"--stdout",
    	"|", bin_path[["bgzip"]], "-c >", output_study_temp,

• I think we should embedded PLINK within the package to avoid having to download a binary. This would make the dependency contained.

[mcanouil]

Note: PLINK is used to convert from VCF to BED/BIM/FAM for flashpcaR to use directly.

In fact, we might want and be able to remove PLINK completely with something like the following:

geno_mat <- unique(data.table::setnames(
  x = data.table::fread(
    cmd = paste(
      bcftools, "query",
      "--format", "'%CHROM\t%POS\t%REF\t%ALT[\t%GT]\n'",
      "--print-header",
      file.path(output_directory, "all-samples-1kg.vcf.gz")
    ),
    sep = "\t"
  ),
  old = function(x) sub("^#", "", sub(":GT", "", sub(" *\\[[[:digit:]]+\\]", "", x)))
))
sample_columns <- setdiff(names(geno_mat), c("CHROM", "POS", "REF", "ALT"))
geno_mat <- geno_mat[
  j = (sample_columns) := lapply(
    X = .SD,
    FUN = function(x) c("00" = 0L, "10" = 1L, "01" = 1L, "11" = 2L)[sub("/|\\|", "", x)]
  ),
  .SDcols = sample_columns
][order(CHROM, POS)]

pca_res <- flashpcaR::flashpca(
  X = geno_mat[j = .SD, .SDcols = !c("CHROM", "POS", "REF", "ALT")],
  ndim = n_comp
)