`credtools prepare`¶

Build LD-backed locus files from chunked summary statistics and genotype references.

credtools prepare INPUTS GENOTYPE_CONFIG OUTPUT_DIR [OPTIONS]

Use prepare when chunk created locus-sized summary-statistics files without LD files, or when you want to rebuild LD from a different reference panel.

Common Use¶

credtools prepare work/chunks/loci_list.txt genotype_config.json work/prepared \
  --threads 4 \
  --ld-format plink

Inputs¶

INPUTS can be a standard CREDTOOLS loci list:

locus_id    chr start   end popu    cohort  sample_size prefix
locus_1 1   50000000    50500000    EUR UKB 400000  work/chunks/EUR.locus_1

It can also be the internal chunk_info.txt shape with ancestry and sumstats_file; prepare normalizes those columns before processing.

GENOTYPE_CONFIG maps each popu value to a genotype prefix. JSON is the simplest form:

{
  "EUR": "/ref/ukb_eur",
  "AFR": "/ref/1kg_afr",
  "EAS": "/ref/1kg_eas"
}

For PLINK references, each prefix must point to .bed, .bim, and .fam files.

Options¶

Option	Meaning	Default
`--threads`	worker count	`1`
`--ld-format`	genotype reference format	`plink`
`--keep-intermediate`	keep temporary PLINK files	off
`--log-file`	write logs to a file	none

VCF extraction is not implemented

The CLI accepts --ld-format vcf, but the current implementation only has a working PLINK extraction path.

Outputs¶

OUTPUT_DIR/
- prepared_files.txt
- loci_list.txt
- {popu}.{locus_id}.sumstats.gz
- {popu}.{locus_id}.ld.npz
- {popu}.{locus_id}.ldmap.gz

Use OUTPUT_DIR/loci_list.txt as the input to qc, meta, finemap, or pipeline.

Relationship to `chunk`¶

If your population config already has ld_ref, chunk runs this preparation step internally and writes work/chunks/loci_list.txt. Running prepare separately is useful when:

you passed direct summary-statistic paths to chunk,
you changed reference panels,
you want a separate, inspectable preparation step before QC and fine-mapping.

credtools prepare¶