Parameters & Options

Comprehensive reference for all command-line parameters in SAIGE-QTL.

Table of contents

• createSparseGRM.R
• step1_fitNULLGLMM_qtl.R
• step2_tests_qtl.R
• step3_gene_pvalue_qtl.R
• makeGroupFile.R

---

createSparseGRM.R

Creates a sparse genetic relationship matrix (GRM). Optional preprocessing step for computational efficiency.

Input Files

Parameter	Type	Default	Description
--plinkFile	character	””	PLINK file prefix (without extension)
--bedFile	character	””	PLINK bed file (if not using --plinkFile)
--bimFile	character	””	PLINK bim file (if not using --plinkFile)
--famFile	character	””	PLINK fam file (if not using --plinkFile)

GRM Options

Parameter	Type	Default	Description
--relatednessCutoff	numeric	0.125	Kinship threshold for relatedness (0.125 = second-degree relatives)
--numRandomMarkerforSparseKin	integer	2000	Number of random markers to identify related samples
--isDiagofKinSetAsOne	logical	FALSE	Set GRM diagonal elements to 1
--minMAFforGRM	numeric	0.01	Minimum MAF for markers used in GRM
--maxMissingRateforGRM	numeric	0.15	Maximum missing rate for markers used in GRM

Computation

Parameter	Type	Default	Description
--nThreads	integer	16	Number of CPU threads
--memoryChunk	numeric	2	Memory chunk size (Gb)
--outputPrefix	character	”~/”	Output path and prefix

---

step1_fitNULLGLMM_qtl.R

Fits the null Poisson mixed model for eQTL analysis.

Package Library

Parameter	Type	Default	Description
--library	character	””	Custom library path for SAIGEQTL package (avoids manual lib.loc specification)

Input Files

Genotype Files

Parameter	Type	Default	Description
--plinkFile	character	””	PLINK file prefix for GRM construction and variance ratio estimation
--bedFile	character	””	PLINK bed file
--bimFile	character	””	PLINK bim file
--famFile	character	””	PLINK fam file

Phenotype Files

Parameter	Type	Default	Description
--phenoFile	character	””	Required. Tab/space-delimited phenotype file with header
--phenoCol	character	””	Required. Phenotype column name (e.g., gene ID)
--sampleIDColinphenoFile	character	“IID”	Required. Sample ID column name
--cellIDColinphenoFile	character	””	Cell ID column name (for single-cell data)
--SampleIDIncludeFile	character	””	File with sample IDs to include

Sparse GRM Files

Parameter	Type	Default	Description
--sparseGRMFile	character	””	Pre-calculated sparse GRM file
--sparseGRMSampleIDFile	character	””	Sample IDs for sparse GRM (no header)

Additional Matrices

Parameter	Type	Default	Description
--VmatFilelist	character	””	Comma-separated list of additional covariance matrices
--VmatSampleFilelist	character	””	Sample IDs for each V matrix

Phenotype and Covariates

Parameter	Type	Default	Description
--traitType	character	“binary”	Required. Trait type: “count” (eQTL), “binary”, or “quantitative”
--covarColList	character	””	Comma-separated covariate columns
--sampleCovarColList	character	””	Comma-separated individual-level covariates
--qCovarColList	character	””	Comma-separated categorical covariates (must also be in --covarColList)
--offsetCol	character	NULL	Offset column (e.g., log total UMI counts)
--varWeightsCol	character	NULL	Variance weights column
--isRemoveZerosinPheno	logical	FALSE	Remove zeros from phenotype
--invNormalize	logical	FALSE	Inverse normalize quantitative traits
--minCovariateCount	numeric	-1	Min count for binary covariates (-1 = no filter)

GRM and Relatedness

Parameter	Type	Default	Description
--useGRMtoFitNULL	logical	TRUE	Use GRM in null model fitting
--useSparseGRMtoFitNULL	logical	FALSE	Use sparse GRM for null model
--useSparseGRMforVarRatio	logical	FALSE	Use sparse GRM for variance ratio estimation
--relatednessCutoff	numeric	0	Kinship threshold for sparse GRM
--isDiagofKinSetAsOne	logical	FALSE	Set GRM diagonal to 1
--minMAFforGRM	numeric	0.01	Min MAF for GRM markers
--maxMissingRateforGRM	numeric	0.15	Max missing rate for GRM markers

Model Fitting

Parameter	Type	Default	Description
--tol	numeric	0.02	Convergence tolerance
--maxiter	integer	20	Maximum iterations
--tolPCG	numeric	1e-5	PCG convergence tolerance
--maxiterPCG	integer	500	Maximum PCG iterations
--tauInit	character	“0,0”	Initial tau values (comma-separated)
--isCovariateOffset	logical	TRUE	Estimate fixed effect coefficients
--isCovariateTransform	logical	TRUE	Apply QR transformation to covariates
--skipModelFitting	logical	FALSE	Skip fitting (requires existing model file)
--solverMethod	character	“auto”	Choose different method to solve the linear system, default being automatic selection based on phenotype data structure, if there are multiple cells data for 1 individual and no GRM is provided, it will automatically use SMW approach. This option can be specified as ‘pcg’ (Preconditioned Conjugate Gradient) or ‘smw’ (Sherman-Morrison-Woodbury)

Variance Ratio Estimation

Parameter	Type	Default	Description
--skipVarianceRatioEstimation	logical	FALSE	Skip variance ratio estimation
--numRandomMarkerforVarianceRatio	integer	30	Initial number of random markers (auto-increases)
--ratioCVcutoff	numeric	0.001	CV threshold for variance ratio
--traceCVcutoff	numeric	0.0025	CV threshold for trace estimation
--nrun	numeric	30	Number of runs for trace estimation
--includeNonautoMarkersforVarRatio	logical	FALSE	Include non-autosomal markers

Categorical Variance Ratios

Parameter	Type	Default	Description
--isCateVarianceRatio	logical	FALSE	Estimate variance ratios by MAC category
--cateVarRatioMinMACVecExclude	character	“10,20.5”	Lower MAC bounds (comma-separated)
--cateVarRatioMaxMACVecInclude	character	“20.5”	Upper MAC bounds (comma-separated)

LOCO (Leave-One-Chromosome-Out)

Parameter	Type	Default	Description
--LOCO	logical	TRUE	Apply LOCO approach
--isLowMemLOCO	logical	FALSE	Output model files by chromosome (reduces memory)

Output Control

Parameter	Type	Default	Description
--outputPrefix	character	”~/”	Required. Output path and prefix
--outputPrefix_varRatio	character	””	Variance ratio file prefix (defaults to --outputPrefix)
--IsOverwriteVarianceRatioFile	logical	FALSE	Overwrite existing variance ratio file
--isStoreSigma	logical	TRUE	Store inverse Sigma matrix
--isShrinkModelOutput	logical	TRUE	Remove unnecessary objects (reduces file size)
--isExportResiduals	logical	FALSE	Export residual vectors

Computation

Parameter	Type	Default	Description
--nThreads	integer	1	Number of CPU threads
--memoryChunk	numeric	2	Memory chunk size (Gb)
--SPAcutoff	numeric	2	SPA threshold (in SD units)

Sex-Specific Analysis

Parameter	Type	Default	Description
--FemaleOnly	logical	FALSE	Run for females only (requires --sexCol, --FemaleCode)
--MaleOnly	logical	FALSE	Run for males only (requires --sexCol, --MaleCode)
--sexCol	character	””	Sex column name
--FemaleCode	character	“1”	Female indicator value
--MaleCode	character	“0”	Male indicator value

---

step2_tests_qtl.R

Performs association tests using the null model from Step 1.

Package Library

Parameter	Type	Default	Description
--library	character	””	Custom library path for SAIGEQTL package (avoids manual lib.loc specification)

Genotype Input Files

Choose one format:

VCF/BCF Files

Parameter	Type	Default	Description
--vcfFile	character	””	VCF/BCF file path
--vcfFileIndex	character	””	VCF index (.csi)
--vcfField	character	“DS”	Field to extract: “DS” (dosage) or “GT” (genotype)

SAV Files

Parameter	Type	Default	Description
--savFile	character	””	SAV file path
--savFileIndex	character	””	SAV index (.s1r)

BGEN Files

Parameter	Type	Default	Description
--bgenFile	character	””	BGEN file (v1.2, 8-bit compression)
--bgenFileIndex	character	””	BGEN index (.bgi)
--sampleFile	character	””	Sample file for BGEN (required for v≥0.38)

PLINK Files

Parameter	Type	Default	Description
--bedFile	character	””	PLINK bed file
--bimFile	character	””	PLINK bim file
--famFile	character	””	PLINK fam file
--AlleleOrder	character	“alt-first”	Allele order: “alt-first” or “ref-first”

Variant Selection

Parameter	Type	Default	Description
--idstoIncludeFile	character	””	File with variant IDs (chr:pos_ref/alt, one per line)
--rangestoIncludeFile	character	””	File with regions (chr, start, end; no header)
--chrom	character	””	Chromosome (required for LOCO or VCF/SAV)

Quality Filters

Parameter	Type	Default	Description
--minMAF	numeric	0	Minimum minor allele frequency
--minMAC	numeric	0.5	Minimum minor allele count
--minInfo	numeric	0	Minimum imputation INFO (if --is_imputed_data=TRUE)
--maxMissing	numeric	0.15	Maximum missing rate
--is_imputed_data	logical	FALSE	Whether data is imputed
--impute_method	character	“best_guess”	Missing imputation: “best_guess”, “mean”, or “minor”
--dosage_zerod_cutoff	numeric	0.2	Dosage zeroing threshold (for imputed data)
--dosage_zerod_MAC_cutoff	numeric	10	MAC threshold for dosage zeroing

Step 1 Model Files

Choose one:

Single Phenotype

Parameter	Type	Default	Description
--GMMATmodelFile	character	””	Model file (.rda) from Step 1
--varianceRatioFile	character	””	Variance ratio file (.varianceRatio.txt) from Step 1

Multiple Phenotypes

Parameter	Type	Default	Description
--GMMATmodel_varianceRatio_multiTraits_File	character	””	3-column file: phenotype, model file, variance ratio file

Single-Variant Tests

Parameter	Type	Default	Description
--LOCO	logical	TRUE	Apply LOCO (requires --chrom)
--SPAcutoff	numeric	2	SPA threshold (in SD units)
--is_Firth_beta	logical	FALSE	Use Firth correction for effect size (binary traits)
--pCutoffforFirth	numeric	0.01	P-value threshold for Firth

Set-Based Tests

Group File

Parameter	Type	Default	Description
--groupFile	character	””	Group file with variant sets

Variant Filters

Parameter	Type	Default	Description
--maxMAF_in_groupTest	character	“0.0001,0.001,0.01”	Max MAF for masks (comma-separated)
--minMAF_in_groupTest_Exclude	character	NULL	Min MAF for masks (comma-separated)
--maxMAC_in_groupTest	character	“0”	Max MAC for masks (comma-separated)
--minMAC_in_groupTest_Exclude	character	NULL	Min MAC for masks (comma-separated)
--annotation_in_groupTest	character	“lof,missense;lof,missense;lof;synonymous”	Annotations (; combines multiple)
--MACCutoff_to_CollapseUltraRare	numeric	10	MAC threshold for collapsing ultra-rare variants
--minGroupMAC_in_BurdenTest	numeric	5	Min MAC for burden test pseudo-marker

Test Types

Parameter	Type	Default	Description
--r.corr	numeric	0	Test type: 0=SKAT-O, 1=Burden only
--is_SKATO	logical	FALSE	Perform SKAT-O test
--weights.beta	character	“1;25”	Beta weights (a;b), multiple sets with commas
--is_equal_weight_in_groupTest	logical	FALSE	Use equal weights
--is_single_in_groupTest	logical	FALSE	Output single-variant results
--is_output_markerList_in_groupTest	logical	FALSE	Output variant lists for each mask

Conditional Analysis

Parameter	Type	Default	Description
--condition	character	””	Conditioning variants (chr:pos_ref/alt, comma-separated)
--weights_for_condition	character	NULL	Weights for conditioning variants (; and , separated)

Sparse GRM

Parameter	Type	Default	Description
--sparseGRMFile	character	””	Sparse GRM file from Step 1
--sparseGRMSampleIDFile	character	””	Sample IDs for sparse GRM
--relatednessCutoff	numeric	0	Kinship threshold
--is_sparseGRM	logical	FALSE	Use sparse GRM

Output

Parameter	Type	Default	Description
--SAIGEOutputFile	character	””	Required. Output file path
--markers_per_chunk	numeric	10000	Variants per chunk (single-variant tests)
--groups_per_chunk	numeric	100	Groups per chunk (set-based tests)
--markers_per_chunk_in_groupTest	numeric	100	Variants per chunk for covariance matrix
--is_output_moreDetails	logical	FALSE	Output genotype counts
--is_overwrite_output	logical	TRUE	Overwrite existing output

Variance Ratio Categories

Parameter	Type	Default	Description
--cateVarRatioMinMACVecExclude	character	“10,20.5”	Lower MAC bounds (must match Step 1)
--cateVarRatioMaxMACVecInclude	character	“20.5”	Upper MAC bounds (must match Step 1)

Advanced Options

Parameter	Type	Default	Description
--is_fastTest	logical	FALSE	Use fast test mode
--pval_cutoff_for_fastTest	numeric	0.05	P-value cutoff for fast mode
--is_noadjCov	logical	TRUE	Regress covariates from genotypes
--max_MAC_for_ER	numeric	4	Max MAC for efficient resampling
--is_EmpSPA	logical	FALSE	Use empirical SPA

---

step3_gene_pvalue_qtl.R

Calculates gene-level p-values using ACAT (Aggregated Cauchy Association Test).

Package Library

Parameter	Type	Default	Description
--library	character	””	Custom library path for SAIGEQTL package (avoids manual lib.loc specification)

Parameter	Type	Default	Description
--assocFile	character	””	Required. Association file from Step 2
--weightFile	character	””	Weight file (2 columns: MarkerID, weight)
--geneName	character	””	Gene name for output (default: “gene”)
--genePval_outputFile	character	””	Required. Output file path

Note: Weight file format requires header with columns “MarkerID” and “weight”.

---

makeGroupFile.R

Creates group files for set-based association tests.

Package Library

Parameter	Type	Default	Description
--library	character	””	Custom library path for SAIGEQTL package (avoids manual lib.loc specification)

Genotype Input Files

Choose one format:

VCF Files

Parameter	Type	Default	Description
--vcfFile	character	””	VCF file path
--vcfFileIndex	character	””	VCF index (.csi)
--vcfField	character	“DS”	Field: “DS” or “GT”

SAV Files

Parameter	Type	Default	Description
--savFile	character	””	SAV file path
--savFileIndex	character	””	SAV index (.s1r)

BGEN Files

Parameter	Type	Default	Description
--bgenFile	character	””	BGEN file (v1.2, 8-bit)
--bgenFileIndex	character	””	BGEN index (.bgi)
--sampleFile	character	””	Sample file for BGEN

PLINK Files

Parameter	Type	Default	Description
--bedFile	character	””	PLINK bed file
--bimFile	character	””	PLINK bim file
--famFile	character	””	PLINK fam file

Region and Output

Parameter	Type	Default	Description
--regionFile	character	””	Required. Regions file (chr, start, end; no header)
--AlleleOrder	character	“alt-first”	Allele order: “alt-first” or “ref-first”
--outputPrefix	character	”~/”	Required. Output path and prefix

---

Required parameters are marked in bold in the Description column. These must be specified for the script to run.

---