Input Files

CREDTOOLS has three main entry points:

1. a population config for raw genome-wide files,
2. a loci list for prepared locus-level files.
3. a genotype config when you run credtools prepare separately.

Use the population config when you want CREDTOOLS to create loci for you. Use the loci list when your locus files already exist. Use the genotype config only when credtools prepare needs to extract LD from PLINK references.

Population Config

Use this with credtools munge and credtools chunk.

popu    cohort  sample_size path    ld_ref
EUR UKBB    400000  /data/eur.sumstats  /ref/EUR
AFR MVP 90000   /data/afr.sumstats  /ref/AFR
EAS BBJ 180000  /data/eas.sumstats  /ref/EAS

Column	Required for	Meaning
popu	munge, chunk	population or ancestry label
cohort	munge, chunk	cohort or study label
sample_size	munge, chunk	sample size for the row
path	munge, chunk	summary statistics path
ld_ref	chunk with LD extraction	PLINK prefix for .bed/.bim/.fam

credtools munge accepts extra columns, so it is fine to include ld_ref from the beginning.

Use stable labels

Keep popu short (EUR, AFR, EAS) and keep cohort readable (UKBB, MVP, BBJ). These labels appear in output files.

Summary Statistics Columns

After munging, CREDTOOLS expects standard columns:

SNPID   CHR BP  RSID    EA  NEA EAF MAF BETA    SE  P   N

The raw input can use common aliases such as CHROM, POS, A1, A2, PVAL, or p_value. credtools munge tries to map them automatically.

If your headers are unusual, create a mapping:

{
  "column_mapping": {
    "chromosome_name": "CHR",
    "genomic_position": "BP",
    "tested_allele": "EA",
    "other_allele": "NEA",
    "effect": "BETA",
    "stderr": "SE",
    "p_value": "P"
  }
}

Then run:

credtools munge population_config.tsv work/munged --config column_mapping.json

Loci List

Use this with credtools meta, credtools qc, credtools finemap, and credtools pipeline.

locus_id    chr start   end popu    cohort  sample_size prefix
locus_1 1   50000000    50500000    EUR UKBB    400000  data/EUR_UKBB_locus_1
locus_1 1   50000000    50500000    AFR MVP 90000   data/AFR_MVP_locus_1

Column	Meaning
locus_id	groups rows from the same genomic region
chr	chromosome
start	locus start position
end	locus end position
popu	population label
cohort	cohort label
sample_size	sample size
prefix	file prefix, without extension

All rows with the same locus_id must have the same chr, start, and end. Each popu + cohort + locus_id combination must be unique.

Genotype Config for prepare

Use a genotype config when chunked summary statistics exist but LD files still need to be extracted from genotype references.

JSON:

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

TSV:

popu    ld_ref
EUR /ref/ukb_eur
AFR /ref/1kg_afr
EAS /ref/1kg_eas

Each key must match a popu value in the input loci list. Each value is a genotype prefix, usually a PLINK prefix without .bed, .bim, or .fam.

Files Behind Each Prefix

CREDTOOLS searches for:

Data	Accepted names
summary statistics	{prefix}.sumstat, {prefix}.sumstats.gz
LD matrix	{prefix}.ld, {prefix}.ld.npz
LD map	{prefix}.ldmap, {prefix}.ldmap.gz

The prefix is not a directory

If prefix is data/EUR_locus_1, CREDTOOLS reads data/EUR_locus_1.sumstats.gz, not data/EUR_locus_1/sumstats.gz.

CLI workflows expect LD files

Even if you plan to use --tool abf, the current CLI loader expects an LD matrix and LD map behind each prefix. Use the Python API if you need a true no-LD ABF run.

Quick Input Checklist

• Paths are relative to the directory where you run the command, or absolute.
• Summary statistics and LD map use the same genome build.
• Alleles are A/C/G/T after munging.
• sample_size is a positive integer.
• prefix does not include an extension.
• The LD matrix row order matches the LD map row order.