Sex Prevalence Adjusted Allelic Difference Estimates Quality Control Metric for GWAS

SPADE

We developed SPADE, an R-based command-line tool for allele frequency quality control and preliminary detection of sex-differentiated genetic signals. SPADE is designed to assess sex-specific allele frequency differences while accounting for disease prevalence and study design.

The tool accepts PLINK-formatted genotype count files as input and supports both pseudo-autosomal (PAR) and non-PAR variants, automatically applying appropriate statistical procedures for each genomic context. SPADE performs input validation, filtering, statistical computation, and comprehensive logging to ensure reproducibility.

Through integration with the argparse R package, SPADE provides a user-friendly command-line interface, allowing users to specify parameters such as case/control sample sizes, population prevalence, and output configurations. The software supports standard command-line options (e.g., --help) and records all runtime arguments.

To facilitate genome-wide analyses, SPADE can be combined with lightweight bash scripting to enable chromosome-level parallelization, allowing efficient large-scale processing.

Publication

Methodological details are available in our preprint:

Controls-Only Quality Control Metric from Early GWAS Can Attenuate Gene-Sex Interaction Signals in Contemporary Large-Scale Studies

https://doi.org/10.64898/2026.07.22.26358697

Citation

If you use SPADE in your research, please cite:

@article{chen2026spade,

  author = {Chen, D. Z. and others},

  title = {Controls-Only Quality Control Metric from Early GWAS Can Attenuate Gene-Sex Interaction Signals in Contemporary Large-Scale Studies},

  journal = {medRxiv},

  year = {2026},

  doi = {10.64898/2026.07.22.26358697}

}

Dependencies

• R (>= 3.5)
• argparse

Install the required R package:

install.packages("argparse")

Input Data Format

SPADE accepts PLINK-formatted genotype count files (.gcount) as input. For each analysis, four input files are required:

• Female cases (.gcount)
• Male cases (.gcount)
• Female controls (.gcount)
• Male controls (.gcount)

Additionally, a PLINK .bim file is required for variant annotation.

Usage

Rscript SPADE.r \
  -fd female_cases.gcount \
  -md male_cases.gcount \
  -fh female_controls.gcount \
  -mh male_controls.gcount \
  -pf 0.04 \
  -pm 0.08 \
  -o output_prefix \
  -b input.bim \
  --mac 5

Command-line Options

Option	Description
-fd	Female cases genotype count file (PLINK .gcount format)
-md	Male cases genotype count file
-fh	Female controls genotype count file
-mh	Male controls genotype count file
-pf	Disease prevalence in females (e.g., 0.04 for 4%)
-pm	Disease prevalence in males
-o	Output prefix for result files
-b	PLINK .bim file for variant annotation
--mac	Minimum allele count threshold for filtering
--no-indels	Indels filter
--help	Show help message

Sample Output

Below is an example of the SPADE output format (first two rows shown):

CHROM	BP	ID	A1	A2	hF_A1A1	hF_A1A2	hF_A2A2	hM_A1A1.A1	hM_A1A2	hM_A2A2.A2	dF_A1A1.A1	dF_A1A2	dF_A2A2.A2	dM_A1A1.A1	dM_A1A2	dM_A2A2.A2	hF_MISSING_CT	dF_MISSING_CT	hM_MISSING_CT	dM_MISSING_CT	EAF_FemaleControls	EAF_MaleControls	EAF_FemaleCases	EAF_MaleCases	EAFsd_FemaleControls	EAFsd_MaleControls	EAFsd_FemaleCases	EAFsd_MaleCases	N_FemaleControls	N_MaleControls	N_FemaleCases	N_MaleCases	EAFDiff_CaseOnly_Sex	EAFDiff_ControlOnly_Sex	EAFDiff_EstPopulation_Sex	EAFDiffsd_EstPopulation_Sex	log10P_EstPopulation_Sex	log10P_CaseOnly_Sex	log10P_ControlOnly_Sex	log10P_FisherCombinedCaseControl
X	1	rs1	A	G	18168	69943	11889	50741	NA	49259	24422	67189	8389	82292	NA	17708	0	0	0	0	0.468605	0.49259	0.419835	0.17708	0.000861143739308369	0.0015809651858912	0.000869488773791818	0.00120715646707459	1e+05	1e+05	1e+05	1e+05	0.242755	-0.023985	0.000695000000000001	0.00167615652382169	0.16851023353268	5784.1054780858	39.7686584656196	Inf
2	2	rs2	A	G	69342	28490	2168	68725	29061	2214	61334	34986	3680	69741	28182	2077	0	0	0	0	0.16413	0.167445	0.21173	0.16168	0.000812196670148308	0.000817035323440792	0.00089126543240496	0.00080675013232103	1e+05	1e+05	1e+05	1e+05	0.05005	-0.00331500000000001	0.000949800000000001	0.0010855384567513	0.418395005039165	378.104713339255	2.3970140241462	Inf

Output

SPADE outputs sex-prevalence-adjusted allele frequency estimates, variance measures, and multiple test statistics for case-only, control-only, and population-level inference.

Output Columns

Column	Description
CHROM	Chromosome
BP	Base pair position
ID	SNP ID
A1	Reference allele
A2	Alternate allele
hF_A1A1, hF_A1A2, hF_A2A2	Female control genotypes
hM_A1A1.A1, hM_A1A2, hM_A2A2.A2	Male control genotypes
dF_A1A1, dF_A1A2, dF_A2A2	Female case genotypes
dM_A1A1.A1, dM_A1A2, dM_A2A2.A2	Male case genotypes
hF_MISSING_CT, hM_MISSING_CT, dF_MISSING_CT, dM_MISSING_CT	Missing counts per group
EAF_FemaleControls, EAF_MaleControls, EAF_FemaleCases, EAF_MaleCases	Estimated allele frequencies
EAFsd_FemaleControls, EAFsd_MaleControls, EAFsd_FemaleCases, EAFsd_MaleCases	SD of allele frequencies
N_FemaleControls, N_MaleControls, N_FemaleCases, N_MaleCases	Sample sizes
EAFDiff_CaseOnly_Sex	Case-only allele frequency difference
EAFDiff_ControlOnly_Sex	Control-only allele frequency difference
EAFDiff_EstPopulation_Sex	Sex-prevalence adjusted allele frequency difference
EAFDiffsd_EstPopulation_Sex	SD of adjusted difference
log10P_EstPopulation_Sex	-log10 p-value of adjusted test
log10P_CaseOnly_Sex	-log10 p-value for case-only comparison
log10P_ControlOnly_Sex	-log10 p-value for control-only comparison
log10P_FisherCombinedCaseControl	Fisher combined p-value

Log File

SPADE automatically generates a log file with the suffix .log containing:

• Runtime arguments and parameter settings
• Input file paths and validation results
• Filtering statistics (e.g., variants removed by MAC threshold)
• Warning and error messages
• Processing timestamps This log ensures reproducibility and aids in debugging.

Features

• Supports both pseudo-autosomal (PAR) and non-PAR variants with appropriate statistical procedures
• Performs input validation, filtering, and comprehensive logging
• Records all runtime arguments for reproducibility
• Can be combined with lightweight bash scripting for chromosome-level parallelization

Example: Parallel Processing Across Chromosomes

for chr in {1..22}; do
  Rscript SPADE.r \
    -fd "chr${chr}_female_cases.gcount" \
    -md "chr${chr}_male_cases.gcount" \
    -fh "chr${chr}_female_controls.gcount" \
    -mh "chr${chr}_male_controls.gcount" \
    -pf 0.04 -pm 0.08 \
    -o "output_chr${chr}" \
    -b "chr${chr}.bim" \
    --mac 500 &
done
wait

Example Command

Rscript SPADE.r -fd toy-female_cases.gcount -md toy-male_cases.gcount -fh toy-female_controls.gcount -mh toy-male_controls.gcount -pf 0.04 -pm 0.08 -o toy_output -b toy.bim --mac 500

SPADEmissing.R: Sex-Dependent Missingness Analysis

A lightweight companion tool to assess differential missingness rates between sexes in case and control groups.

Usage

Rscript SPADEmissing.R input.txt output.txt

Description

SPADEmissing.R takes a single result file from SPADE (formatted as shown in the Sample Output section) and computes:

• OR_Missing_Case_FemaleVsMale: Odds ratio of missingness in female cases vs male cases
• log10P_Missing_Case_FemaleVsMale: -log10(p-value) for case group sex difference in missingness
• OR_Missing_Control_FemaleVsMale: Odds ratio of missingness in female controls vs male controls
• log10P_Missing_Control_FemaleVsMale: -log10(p-value) for control group sex difference in missingness These four columns are appended to the original input file.

Notes

• Missingness is defined as the proportion of samples failing genotype calling at a given variant
• Odds ratio > 1 indicates higher missingness in females; < 1 indicates higher missingness in males
• Fisher's exact test is used for p-value calculation
• For contingency tables with zero counts, a continuity correction of 0.5 is applied