Set YAML file parameters

The cloned repository will contain a params.yml file where the parameters for the analysis can be set. Detailed descriptions of each parameter are present in the Commandline parameters section. There are five main sections to modify:

1. Input files

Specify the path for the input, covarfile, and phenofile files required for the analysis. For example:

# Input files
input: ./longitudinal-GWAS-pipeline/example/genotype/chr[1-3].vcf
covarfile: ./longitudinal-GWAS-pipeline/example/covariates.tsv
phenofile: ./example/phenotype.surv.tsv

See Input file formats for more details on which variables must be present in the phenofile and covarfile.

2. Variable names

Specify the variable names to be used in the analysis as they appear in the input files.

  • pheno_name: the name of the independent variable, i.e. the outcome of interest

  • covariates: the covariates to be used, written in double quotes ("") and separated by spaces

  • study_col: the column in the covarfile where the study arm is specified (e.g. affected/unaffected/intermediate)

  • time_col: the column in the phenofile where the time-to-event is defined

For example, for a survival analysis using "Sex" and "Age at baseline" as covariates, we would specify the following:

#Variables names
pheno_name: surv_y
covariates: "SEX age_at_baseline"
study_col: study_arm
time_col: study_days

3. Model variables

These boolean variables will be used to determine which type of model will be run. Currently, there are three options to choose from: glm, lmm_gallop, and cph. Thus, we can select which model to use by setting the corresponding params.yml parameter to true or false. For example, for a survival analysis we would set the survival_flag to true and the rest to false:

# Model variables
longitudinal_flag: false
survival_flag: true
linear_flag: false

The chunk_flag can be set to true if we wish to divide the genetic files into smaller chunks which can be run in parallel for increased speed of analysis. The chunk size can be specified using the chunk_size and plink_chunk_size parameters:

# Chunk size
chunk_flag: true
chunk_size:
plink_chunk_size:

4. Parameters for genetic QC

Here we specify the parameters for genetic QC, such as maf, ancestry, and assembly. For example:

# Parameters for genetic QC
r2thres: -9
minor_allele_freq: "0.05"
minor_allele_ct: "20"
kinship: "0.177"
ancestry: "EUR"
assembly: "hg19"

Detailed descriptions of each parameter can be found in the Commandline parameters section.

5. Dataset name

Finally, an identifier name can be given to the analysis, which will be used to name the cache and results folders. Additionally, the mh_plot flag can be set to true if we desire to produce Manhattan Plots to visualise the results.

#Identifier for the input genotype files - useful to cache results
dataset: "V2_SURV"
# Generate manhattan with result files
mh_plot: true