Command line Parameters

Users can specify the following parameters when running the longGWAS Nextflow pipeline:

--ancestry

This parameter is used to specify the ancestry group to perform the analysis. Valid values for this parameter are ['EUR', 'SAS', 'EAS', 'AFR', ...]

--assembly

This parameter is used to specify the genome assembly of the input genotyping files. The outputs of the pipeline will all be in relation to hg38. If you supply genotypes from a different reference assembly specify one of the following options ['hg18', 'hg19'] so the positions can be lifted over to hg38. The default value is 'hg38'

--chunk_flag

Set to true to divide the genetic files into smaller chunks (specified by --chunk-size), which allows us to run the QC steps in parallel for increased efficiency.

--chunk_size

Specify the chunk size used to split the genetic files for genetic QC. The default value 30000, meaning the genetic files will be split into chunks of 30,000 SNPs.

--covarfile

This parameter is used to specify the path of the covariates to include in the model. Each subject to be included in the analysis needs to have their own covariates. For more details, see the page on file inputs to the pipeline.

--covariates

This parameter is used to specify the covariates to include in the model from the input covariates file and the genetic principle components from the ancestry steps. This parameter should be populated with a space-delimited string of the column names to include in the model from the --covarfile option. By default, the model is fit with "SEX PC1 PC2 PC3"

--dataset

This parameter is used to specify an identifier for the input genotype files so that subsequent re-runs of the pipeline can use cached results from the variant filtering and ancestry + outlier detection steps. The default value is null (''), and it is highly recommended to set this parameter to avoid misusing cached results.

--input_vcf

This parameter is used to specify the paths of the genotyping files. The pipeline requires genotypes in uncompressed (*.vcf) or compressed (*.vcf.gz) files. Genotype files should be shredded at the chromosome level and each file should contain the chromosome number prefix with 'chr' case insensitive.

Note: The inclusion of wildcard (*) in the path requires the use of quotes

Acceptable filenames

--input_vcf "/path/to/dataset_prefix_chr*.vcf"
--input_vcf "/path/to/dataset_chr*_suffix.vcf.gz"
--input_vcf "/path/to/chr*.vcf"
--input_vcf "/path/to/*chr[1-5].vcf"

--kinship

This parameter specifies the relatedness level to filter against from the pairwise kinship between subjects. By default this value is "0.177" to filter out first-degree relations.

--longitudinal_flag

This flag is used to specify the estimation of longitudinal associations via the GALLOP algorithm. If the option is not supplied then the pipeline performs a cross-sectional GWAS using plink2.

Note: performing longitudinal analysis requires the inclusion of a study_days variable in the input phenotype. This variable should correspond to the timepoint (in days) since the start of the study at which the measurement for the outcome was taken. Initial measurements taken at baseline will have a study_days of 0.

--mh_plot

Set to true to produce a Manhattan Plot to visualise the results of the analysis, as well as a QQ plot for diagnostic purposes.

--minor_allele_freq

Set the maf threshold to filter out variants. For example, to filter out variants with maf<5%:

minor_allele_freq: "0.05"

--minor_allele_ct

Set the mac threshold to filter out variants. For example, to filter out variants with mac<"20":

minor_allele_ct: "20"

--model

This parameter can be used to specify a custom model with higher order terms when the --longitudinal_flag is invoked. To include higher-order terms in the cross-sectional analysis, include them as a column in the --covarfile and declare them in the --covariates parameter.

--out

This parameter is used to specify the output suffix of the files to distinguish results or re-runs of the pipeline.

--phenofile

This parameter is used to specify the path of the outcome file to do the association on. For cross-sectional analysis, the pipeline expects data to be formatted in at least 2 columns, IID and y.

For longitudinal analysis, a third column is required containing information about the time point in the study that the observation was taken. More specifically, this variable should represent the number of days since the starting point of your analysis. The name of this variable should be passed using the --time-col flag.

--pheno_name

This parameter can be used to specify the column in the --phenofile containing the outcome of interest. Additionally, longGWAS enables running the analysis using multiple outcomes simultaneously. If multiple outcomes are present, they must be passed in a comma-separated manner y,y_2,z,z_2.

--r2thres

This parameter is used to filter out imputed genotypes of low quality if the input genotyping files include imputed variants.

--study_col

Specify the name of the study_arm column in the --covarfile.

--time_col

Used for survival analyses: specify the column name in the --phenofile which contains the time-to-event information in days.