.fam Fileset to .rasbin
In RAS: Regional Association Score for Genome-Wide Association Studies

bed_to_rasbin

R Documentation

Convert a PLINK 1 .bed/.bim/.fam Fileset to .rasbin

Description

Converts a PLINK 1 binary fileset directly into the chunked .rasbin format, without ever materializing a dense genotype matrix in R. Both the read from .bed and the write to .rasbin are done in fixed-size SNP chunks in C (see src/plink_io.c), so peak memory is O(\text{n\_samples} \times \text{chunk\_snps}) regardless of how many samples or SNPs the fileset contains.

Usage

bed_to_rasbin(bed_path, out_path, chunk_snps = 2000)

Arguments

`bed_path`	Character. Path to the `.bed` file. The companion `.bim` and `.fam` files are expected alongside it with the same stem (standard PLINK convention, e.g. `--bfile` prefix).
`out_path`	Character. Destination path for the `.rasbin` file. A companion `<out_path>.meta.rds` is written alongside it holding sample and SNP identifiers, parsed from `.fam`/`.bim`.
`chunk_snps`	Integer. Number of SNPs read from `.bed` and written to `.rasbin` per chunk. Only affects peak memory and throughput, not the resulting file. Default `2000`, matching `geno_to_rasbin`'s `chunk_write_size` default.

Details

Only SNP-major .bed files are supported (the only mode any current PLINK version writes). Genotype dosage counts copies of the A1 allele (the fourth column of .bim), matching PLINK's own --recode A / plink2's --export A convention: 2 = homozygous A1, 1 = heterozygous, 0 = homozygous A2, NA = missing.

Unlike geno_to_rasbin (dtype 0, double, 8 bytes/genotype), this writes dtype 1 (int8, 1 byte/genotype) – 8x smaller on disk, though still ~4x larger than .bed's 2-bit packing (see src/rasbin.h for why byte-per-genotype was chosen over bit-packing). Both dtypes are read transparently by rasbin_read_chunk and every ⁠_fast⁠ pipeline function.