• matrix_eqtl_wrapper.py

  • The main entrypoint for the pipeline.
  • Takes four arguments
    • -v/--vcf-file : name of VCF file to use in eQTL analysis
    • -g/--gene-expression-file : name of gene expression matrix to use in eQTL analysis
    • -p/--gene-position-file : name of file that contains the positions of the genes
    • -n/--numfactors : number of PEER factor to correct for
  • Has 28 optional arguments
    • --headless-vcf-filename : Custom name of output file if the .noh extension isn't desired
    • --overlap-extension : Custom file extension to be used if .out isn't desired
    • --maf-cutoff : MAF cutoff to use. Default is 0.05
    • --filtered-filename: Custom name of output file if .maf_filtered isn't desired
    • --parsed-filename : Custom name of output file if .matrix isn't desired
    • --position-filename : Custom filename for the genotype positions
    • --meqtl-position-filename : Custom filename for the MatrixEQTL formatted positions
    • --number-pcs : Number of PCs to include; default=1
    • --pc-filename : Custom filename for the PC covariates, default is filename.pcs
    • --normalized-filename : Custom filename for the inverse quantile normalized expression matrix, default is filename.qnorm
    • --peer-factor-filename : Custom filename for the PEER factors, default is filename.peer_factors_n where n is the number of factors
    • --combined-covariate-filename : Custom filename for the combined covariates, default is combined_covariates
    • --additional-covariates : Filename for any additional covariates to include in the combined covariates file
    • --trans-output-file : Filename of the output file for trans association, default is no trans- output filename
    • --trans-p-value : P-value cutoff for trans associations, default is 0
    • --model: Model type to use. Default is linear
    • --cis-distance : Maximum distance between genotype and gene
    • --cis-p-value : P-value cutoff, default is 0.05
    • --no-header : Binary flag to use if matrices don't have a header with IDs
    • --no-rownames : Binary flag to use if matrices don't have rowname IDs
    • --missing : The value of missing data, default is NA
    • --sep : The separating character of the genotype and gene expression matrices, default is "\t"
    • --maf : MAF cutoff for filtering by MatrixEQTL; Default is 0 and no MAF filtering
    • --qqplot : Filename of the qq-plot, default is no qq-plot file (?)
    • --eqtl-output-file : Filename of the output file, default is MatrixEQTLOutput
    • --boxplot-pdf-file : Filename for PDF of boxplots, default is no boxplots
    • --correlation-output-file : Filename for the correlations of the eQTLs. Default is MatrixEQTLOutput.corr
    • --manhattan-pdf-file : Filename for the PDF of the manhattan plot, default is no manhattan plot

• remove_vcf_header.py

  • Removes ## header lines from VCF files; Keeps # header
  • Takes one argument
    • -v/--vcf-file : name of VCF file to remove header from
  • Has two optional arguments
    • -o/--output-file : Custom name of output file if the .noh extension isn't desired
    • -s/--stdout : Prints output to stdout ie console/terminal, can be redirected or piped

• vcf_overlap.py

  • Overlaps the samples from the VCF file and the gene expression file
  • MatrixEQTL requires the VCF file and gene expression file to have the same samples and in the same order
  • Takes two arguments
    • -v/--vcf-file : A VCF file with the header information removed (output of remove_vcf_header.py)
    • -g/--gene-expression-file : A gene expression matrix; Each column is a sample, each row is a gene
  • Takes one optional argument
    • -e/--extension : Custom file extension to be used if .out isn't desired
  • vcf_overlap.py doesn't have a stdout option as it outputs two files

• filter_snps.py

  • Minor allele frequency filtering
  • Takes one argument
    • -v/--vcf-file : A VCF file with the header information removed; N.B. To get accurate MAF filtering for your sample, use this step after overlapping
  • Takes two optional argument
    • -o/--output-file : Custom name of output file if .maf_filtered isn't desired
    • -s/--stdout : Prints output to stdout, so it can be redirected or piped

• parse.py

  • Parses a ## headerless VCF file to get summed genotypes; ex 0|0: becomes 0, 0|1: becomes 1, 1|1: becomes 2
  • Takes one argument
    • -v/--vcf-file : A VCF file with the header information removed
  • Takes two optional arguments
    • -o/--output-file : Custom name of output file if .matrix isn't desired
    • -s/--stdout : Prints output to stdout

• position.py

  • Creates two position files, one in the form required for MatrixEQTL, and one that contains id, chr, start, stop, and type
  • Takes one argument
    • -v/--vcf-file : A VCF file with the header information removed
  • Takes two optional arguments
    • -o/--output-file : Custom filename for the genotype positions
    • -m/--meqtl-file : Custom filename for the MatrixEQTL formatted positions

• pc_covariates.py

  • Calculates Principle Components using Scikit-Learn's IncrementalPCA method
  • Takes one argument
    • -v/--vcf-file : A VCF file that has been parsed with parse.py
  • Takes three optional arguments
    • -o/--output-file : Custom filename for the PC covariates, default is filename.pcs
    • -n/--number-pcs : Number of PCs to include; default=1
    • -s/--stdout : Prints output to stdout

• run_iqn.py

  • Python wrapper for general_iqn_py.R
  • Performs inverse quantile normalization on a gene expression matrix
  • Inverse quantile normalization casts each row (gene expression values from one gene for each sample) onto the standard normal distribution by rank
  • Takes one argument
    • -i/--input-file : A gene expression matrix file; N.B. for the Inverse Quantile Normalization to get the correct results for an eQTL analysis use this step after overlapping
  • Takes two optional arguments
    • -o/--output-file : Custom filename for the inverse quantile normalized expression matrix, default is filename.qnorm
    • -s/--stdout : Prints output to stdout

• run_peer.py

  • Python wrapper for peer_function.R
  • Runs PEER on gene expression matrix to calculate PEER factors which can be used as covariates
  • Takes two arguments
    • -i/--input-file : The gene expression matrix filename; N.B. run on the inverse quantile normalized gene expression matrix
    • -n/--number-factors : The number of PEER factors to calculate
  • Takes one optional argument
    • -o/--output-file : Custom filename for the PEER factors, default is filename.peer_factors_n where n is the number of factors

• combine_covariates.py

  • Combines genotype PC covariates, gene expression PEER factors and any additional PCs like population and gender
  • Takes two arguments
    • -p/--snp-pc-file :
    • -f/--peer-factor-file :
  • Takes three optional arguments
    • -o/--output-file : Custom filename for the combined covariates, default is combined_covariates
    • -a/--additional-covariates : Filename for any additional covariates to include in the combined covariates file
    • -s/--stdout : Prints output to stdout

• run_matrix_eqtl.py

  • Wrapper for running MatrixEQTL

  • Takes four arguments

    • -m/--genotype-matrix : Genotype matrix of summed genotypes
    • -p/--genotype-positions : Genotype positions in MatrixEQTL format
    • -e/--gene-expression-matrix : Gene expression matrix
    • -g/--gene-positions : Gene positions in MatrixEQTL format

  • Takes 13 optional arguments

    • -c/--covariates : Covariate filename, no default
    • -o/--output-file : Custom filename for the MatrixEQTL output, default is MatrixEqtlOutput
    • -v/--p-value : P-value cutoff, default is 0.05
    • -q/--qq-plot : Custom filename for the qq-plot PDF file, default is MatrixEqtlQQPlot.pdf
    • --trans-output-file : Filename of the output file for trans association, default is no trans- output filename
    • --trans-p-value : P-value cutoff for trans associations, default is 0
    • --model : Model type to use. Default is linear
    • --cis-distance : Maximum distance between genotype and gene
    • --maf : MAF cutoff for filtering by MatrixEQTL; Default is 0 and no MAF filtering
    • --no-header : Binary flag to use if matrices don't have a header with IDs
    • --no-rownames: Binary flag to use if matrices don't have rowname IDs
    • --missing : The value of missing data, default is NA
    • --sep : The separating character of the genotype and gene expression matrices, default is "\t"

• CorrBoxPlot.py

  • Code to produce a boxplot for each eQTL
  • Takes three arguments
    • -m/--matrix-eqtl-results : Filename of the eQTL results
    • -g/--genotype-file : File the genotype matrix used to produce the eQTLs
    • -e/--gene-expression-file : Filename of the gene expression matrix to produce the eQTLs
  • Takes three optional arguments
    • -s/--stdout : Option to print the eQTL results with the correlation to stdout
    • -o/--output-file : Name of file for eQTL results with correlation. Default is .corr
    • -p/--pdf-file : Name of PDF file for the boxplots. If not provided boxplots aren't produced.

• manhattan.py

  • Code to create a Manhattan plot from the eQTL results
  • Takes three arguments
    • -e/--eqtl-output-file : Filename of the eQTL results
    • -l/--position-file : Filename for the genotype positions
    • -p/--pdf-file : Filename for the PDF output of the Manhattan plot

• modify_matrix_eqtl.py

  • Script to create a version of MatrixEQTL Engine that behaves better with Python and rpy2
  • Takes two optional arguments
    • -o/--output-file
    • -s/--stdout

---

• general_iqn_py.R

  • Inverse quantile normalization function
  • Takes one argument : Filename of the gene expression matrix to normalize
  • Returns the inverse quantile normalized genen expression matrix

• peer_function.R

  • R function to run PEER
  • Takes two arguments :
    • Filename of gene expression matrix
    • Number of PEER factors to calculate
  • Returns the PEER factor matrix

• mxeqtl.R

  • R functions to run MatrixEQTL
  • The main function (mxeqtl) takes six arguments
    • snp_file : Filename of genotype matrix
    • snp_location : Filename of the genotype positions
    • expr_file : Filename of the gene expression matrix
    • expr_location : Filename of the gene positions
    • cis_output_file : Filename of the output file
    • cis_pval : P-value cutoff
  • Takes 11 optional arguments
    • covariates : Filename of the covariates, default is no covariate file
    • trans_pval : P-value cutoff for trans associations, default is 0
    • trans_output_file : Filename of the output file for trans association, default is no trans- output filename
    • model : Model type to use. Default is linear
    • MAF : MAF cutoff for filtering; Default is 0 and no MAF filtering
    • cis_dist : Maximum distance between genotype and gene
    • qq : Filename of the qq-plot, default is no qq-plot file
    • missing : The value of missing data, default is NA
    • sep : The separating character of the genotype and gene expression matrices, default is "\t"
    • header : Binary flag if matrices have a header with IDs, default is TRUE
    • rownames : Binary flag if matrices have rowname IDs, default is TRUE