Estimate RAM required for a VCFArrow export operation

Usage

vcf_memory_estimate(
  vcf_arrow,
  keep_groups = NULL,
  format = c("individual", "pop", "chunk"),
  chunk_size = 100000L,
  lowmem = FALSE
)

Arguments

vcf_arrow: A VCFArrow object.
keep_groups: Groups to export (NULL = all).
format: One of "individual" (Structure, Arlequin, FASTA, ...) or "pop" (BayesScan, Treemix, Migrate-N C, ...) or "chunk" (SmartSNP, fineRADstructure, sNMF, EIGENSTRAT, ...).
chunk_size: Feather chunk size used at read_vcf() time.
lowmem: If TRUE, estimate uses raw-byte matrices (vcf2*() lowmem variants); otherwise integer matrices.

Value

Invisibly, a list with elements:

n_var: number of variants
n_samples: number of samples retained
n_pops: number of populations
chunk_arrow_bytes: Arrow memory per chunk read
matrix_bytes: size of the accumulation matrices
peak_bytes: estimated peak memory

Examples

f <- system.file("extdata", "vaillantii_discosnp_sub.vcf.gz",
                 package = "VCFArrow")
vcf <- read_vcf(f) |> set_vcf_groups(dirname(f))
#> ℹ VCF is being read in chunks of 50000 variants
#> ✔ VCF successfully read into a VCFArrow object
vcf_memory_estimate(vcf, format = "individual")
#> 
#> ── VCFArrow memory estimate ──
#> 
#> • Variants (filtered): 9313
#> • Samples retained: 18
#> • Populations: 4
#> • Chunk size: 100,000 variants
#> • Low-memory mode: FALSE
#> 
#> ── Per-operation footprint 
#> • Arrow pool per chunk read: 34.3 MiB
#> • Accumulation matrices (a1+a2): 1.3 MiB [4 B/cell]
#> • Estimated peak RAM: 35.6 MiB