binx threshold

Calculate significance thresholds for GWAS results.

Synopsis

binx threshold --results <FILE> --method <METHOD> [OPTIONS]

The threshold command calculates significance thresholds for GWAS using various multiple testing correction methods. It accounts for the number of tests performed and optionally the correlation structure among markers.

Recommendation: Use m.eff for the most accurate threshold as it accounts for linkage disequilibrium (LD) between markers. For large datasets where speed is a concern, bonferroni or fdr are faster alternatives that don’t require genotype data.

Required Arguments

Argument	Description
`--results <FILE>`	GWAS results CSV file
`--method <METHOD>`	Threshold method: bonferroni, m.eff, or fdr

Options

Option	Default	Description
`--alpha <FLOAT>`	0.05	Significance level
`--geno <FILE>`	-	Genotype file (required for m.eff)
`--ploidy <INT>`	-	Ploidy level (required for m.eff)

Methods

Bonferroni

The most conservative approach. Fast for large datasets as it doesn’t require genotype data:

threshold = α / n_tests

binx threshold \
  --results gwas.csv \
  --method bonferroni \
  --alpha 0.05

Effective Number of Tests (M.eff) — Recommended

Accounts for LD between markers using the method of Moskvina & Schmidt (2008). Requires genotype data to calculate marker correlations. This is the most accurate method as it adjusts for correlation between markers.

binx threshold \
  --results gwas.csv \
  --method m.eff \
  --geno geno.tsv \
  --ploidy 4 \
  --alpha 0.05

FDR (Benjamini-Hochberg)

False Discovery Rate control. Fast for large datasets as it doesn’t require genotype data:

binx threshold \
  --results gwas.csv \
  --method fdr \
  --alpha 0.05

Output

Method: Bonferroni
Alpha: 0.05
Number of tests: 50000
P-value threshold: 1.00e-06
-log10(p) threshold: 6.00

Examples

Compare Methods

# Bonferroni and FDR (don't require genotype data)
for method in bonferroni fdr; do
  echo "=== $method ==="
  binx threshold --results gwas.csv --method $method --alpha 0.05
done

# M.eff requires genotype data
binx threshold --results gwas.csv --method m.eff --geno geno.tsv --ploidy 4

Integrated GWAS Workflow (Recommended)

The recommended approach is to use binx gwas --threshold to calculate thresholds during GWAS, which adds the threshold to each result row for use with binx qtl:

# Run GWAS with threshold calculation
binx gwas \
  --geno geno.tsv \
  --pheno pheno.csv \
  --trait yield \
  --ploidy 4 \
  --threshold m.eff \
  --out results.csv

# Extract QTLs (uses threshold column from results)
binx qtl --input results.csv --bp-window 1000000 --output qtls.csv

Adding Thresholds to Existing Results

If you have GWAS results without thresholds, you can calculate them separately and add them to the results file.

Step 1: Calculate thresholds

binx threshold --results gwas_results.csv --method bonferroni

Output:

Thresholds (Bonferroni):
Model                   Threshold       M.eff   n_markers
------------------------------------------------------------
additive                     5.30           -        1000
general                      5.60           -        1000

Step 2: Add threshold column matching by model

# Define thresholds per model (from Step 1 output)
awk -F',' -v OFS=',' '
BEGIN {
    thresh["additive"] = 5.30
    thresh["general"] = 5.60
}
NR==1 { print $0",threshold" }
NR>1 {
    model = $4
    t = (model in thresh) ? thresh[model] : "NA"
    print $0","t
}' gwas_results.csv > gwas_with_threshold.csv

Step 3: Extract QTLs

binx qtl --input gwas_with_threshold.csv --bp-window 1000000 --output qtls.csv

Binx Documentation