Skip to contents

Read Genotype Data into an LDxBlocks Backend

Source: R/read_geno.R
read_geno.Rd

Reads genotype data from disk (or accepts an in-memory matrix) and returns a unified backend object that all LDxBlocks functions can consume. The format is auto-detected from the file extension unless overridden with format.

Usage

read_geno(
  path,
  format = NULL,
  snp_info = NULL,
  sample_ids = NULL,
  sep = ",",
  na_strings = c("NA", "N", "NN", "./.", ".", ""),
  gds_cache = NULL,
  clean_malformed = FALSE,
  verbose = FALSE
)

Arguments

path

Character. Path to the genotype file, OR an R numeric matrix when format = "matrix".

format

Character or NULL. One of "numeric", "hapmap", "vcf", "gds", "bed", "matrix". NULL (default) auto-detects from extension.

snp_info

Data frame with columns SNP, CHR, POS (and optionally REF, ALT). Required only when format = "matrix".

sample_ids

Character vector. Override sample IDs extracted from the file. Length must equal number of samples.

sep

Character. Field separator for "numeric" format.

na_strings

Character vector. Strings treated as NA. Default c("NA", "N", "NN", "./.", ".", "").

gds_cache

Character or NULL. Path where a GDS cache file should be written when format = "vcf" and SNPRelate is available. If NULL (default), the GDS file is placed next to the VCF with a .gds extension. Set to FALSE to disable auto-conversion and read the VCF fully into memory instead. When the cache file already exists it is reused without re-converting (fast subsequent calls). Ignored for all non-VCF formats. Default ",".

clean_malformed

Logical. If TRUE, the input file is stream-cleaned before reading: any line whose delimiter-separated column count does not match the header is silently removed. This adds one extra streaming pass over the file but is required for files produced by some variant callers (e.g. NGSEP) that embed extra characters in FORMAT fields. Applies to numeric dosage CSV, HapMap, and VCF formats. Default FALSE.

verbose

Logical. Print progress messages.

Value

An object of class "LDxBlocks_backend" with elements: type, n_samples, n_snps, sample_ids, snp_info (data.frame SNP/CHR/POS/REF/ALT). Use read_chunk to extract genotype slices and close_backend to release file handles.

Supported formats

"numeric"

CSV or TXT, one row per SNP. Required columns: SNP, CHR, POS, REF, ALT. Remaining columns are sample dosages in {0, 1, 2, NA}. Extension: .csv, .txt.

"hapmap"

Standard 11-column HapMap header followed by sample columns with two-character nucleotide calls (AA, AT, NN for missing). Extension: .hmp.txt.

"vcf"

VCF v4.2. Both phased (0|1) and unphased (0/1) GT fields are accepted. Multi-allelic sites use first ALT. Missing (./.) becomes NA. Extension: .vcf, .vcf.gz.

"gds"

SNPRelate GDS file. Requires BiocManager::install("SNPRelate"). Extension: .gds.

"bed"

PLINK binary BED. Companion .bim and .fam files must exist at the same path stem. Extension: .bed.

"matrix"

In-memory R numeric matrix (individuals x SNPs, 0/1/2). Must supply snp_info separately.

See also

read_chunk, close_backend, run_Big_LD_all_chr

Examples

# \donttest{
# In-memory matrix (existing workflow, unchanged)
G   <- matrix(sample(0:2, 100*500, replace=TRUE), 100, 500)
rownames(G) <- paste0("ind", 1:100)
colnames(G) <- paste0("rs",  1:500)
info <- data.frame(SNP=colnames(G), CHR=rep(1:5, each=100),
                   POS=rep(seq(1e4,5e6,length.out=100),5))
be <- read_geno(G, format="matrix", snp_info=info)
be$n_snps   # 500
#> [1] 500
mat <- read_chunk(be, 1:20)
dim(mat)    # 100 x 20
#> [1] 100  20
close_backend(be)
# }