File Conversion

Related options [top]

WISARD provides several options for file conversion and their combination provides input file formats for several software.

Available options

• --makeraw : recode genotypes as 0(major homo), 1(hetero) and 2(minor homo).
• --outphenoonly : delete columns for FID, IID, paternal ID, maternal ID and SEX.
• --outnoheader : delete headers.
• --out1case : recode case and control status as 1 and 0, respectively.

File Conversion [top]

WISARD can generate the file with the following formats:

• BOOST : suggested by Wan et al. Am J Hum Genet 2010 and it provides relatively fast way to investigate GWAS level gene-gene interaction.
Convert 'sample.bed' into an input of BOOST
C:\Users\WISARD> wisard --bed sample.bed --makeraw --outphenoonly --out1case --outnoheader --out inp_boost
Running the above code generates [prefix].raw file and it can be directly used as an input for BOOST. with below constraints.

NOTE!

Since BOOST requires case/control dataset, this conversion process will not work for the dataset with continuous phenotype

• Beagle : Imputation and genetic analyseis software
was suggested by (Browning et al. Am J Hum Genet 2006) and is used for imputation of missing and untyped genotypes.
Convert 'sample.bed' into an input of Beagle
C:\Users\WISARD> wisard --bed sample.bed --makebeagle --out inp_beagle
• MERLIN : One of pedigree analysis package
was suggested by Abecasis et al.(2002) and it is used for family-based analysis such as haplotyping, linkage analysis, etc.
Convert 'sample.bed' into an input of MERLIN
C:\Users\WISARD> wisard --bed sample.bed --makemerlin --out inp_merlin
• IMPUTE2/SNPTEST: Both softwares use the same input file format. IMPUTE2 was suggeted by Howie et al.(2009) and is used for imputation of missing and untyped genotypes SNPTEST was suggetsed by Marchini et al.(2009) and is used for genome-wide association studdies.
  Convert 'sample.bed' into a binarized input of IMPUTE2/SNPTEST

  C:\Users\WISARD> wisard --bed sample.bed --makebgen --out inp_merlin
  In addition, to default binarization, its genotype data can be unzipped with gzip algorithm as follows:
  Convert 'sample.bed' into a binarized input with zipped data of IMPUTE2/SNPTEST

  C:\Users\WISARD> wisard --bed sample.bed --makebgen --zipbgen --out inp_merlin

Useful Options

Analysis summary [top]

Remove NAs from the output file

If certin information is missing or misspecified, p-values cannot be calculated and NAs are reported. If you want to remove NAs in p-values, --remana can be used. In order to remove those NA p-values from the result of analyses, --remna can be used.

If multiple p-values for a single variant/gene are reported, result for a variant/gene are reported unless all p-values are NA.

Example code

Filter variants of which p-values are in certain range

Results of variants/genes where their p-values are in certain range can be selectively reported by using --pvalrange option. If multiple p-values for a single variant/gene are reported, result for a variant/gene are reported if at least a signe p-value is in the specified range.

Example code

Report computational time for each analysis

Computational time for most of analysis supported by WISARD can be reported by using --time option.

Example code

Annotation summary [top]

Annotation for association analyses provides various informations about the disease susceptibility locus. WISARD provides some functions for annotation by using information of genetic/physical position for each variant.

Annotate variants with each gene

Result for each variant within a gene can be annotated by using --annogene option and they will be provided in the additional column, ANNO.

Example code

Expading annotation range

WISARD provides further function for annotating variants which are located near genes but not within genes. "--annorange k" makes WISARD annotates all variants of which distances with adjacent gene are less than k basepair.

Analyses supporting annotation functions

For the following analyses with WISARD, functions for annotating each variants can be applied:

• Regression analysis
• Quasi-likelihood-based analysis such as MQLS and FQLS
• MAF estimation
• HWE estimation