RAREMETALWORKER SPECIAL TOPICS

This page describes how RAREMETALWORKER handles some special cased during analyses.

Useful Links

Here are some useful links to key pages:

• RAREMETALWORKER documentation
• RAREMETALWORKER method
• RAREMETALWORKER command reference
• FAQ

Unrelated Individuals

RAREMETALWORKER generates single variant association test statistics for a single study prior to meta-analysis. This page provides a brief description of the statistics that RAREMETALWORKER calculates, together with key formulae.

Missing Data

• Individuals with missing phenotypes will be excluded from analysis.
• If --makeResiduals is used for adjusting covariates, then individuals with missing covariates will also be excluded.
• Individuals that are not genotyped will also be excluded from analyses.
• Missing genotypes are imputed using mean genotype of a variant.

Analyzing Chromosome X

• To make sure RAREMETALWORKER analyze chromosome X correctly, male must be code as 1 and female must be coded as 2 in PED file.
• --xStart and --xEnd options described the start and end position of nonPAR region on chromosome X. The default values are 2699520 and 154931044, according to Human Genome build 19.
• When samples are analyzed as unrelated (no any kind of kinship is used for linear mixed model approach), no special actions needed for analyzing markers on chromosome X.
• If you want to use a linear mixed model approach to correct family structure, population structure, or cryptic relatedness, you should issue --vcX option in your command line. This approach will fit a separate linear mixed model using both autosomal kinship matrix and chromosomeX kinship matrix. This approach is believed to have larger power with type I error well controlled.
• On the other hand, if you want a fast association for chromosome X, another valid approach is to issue --vcX --separateX options. This will initiate fitting a separate linear mixed model using chromosomeX kinship only for analyzing markers on chromosomeX. This approach is expected to be faster than the above option, but with less power, although type I error is also expected to be under control.
• Male genotypes for variants in nonPAR region are coded to be 0 or 2.
• If a male is heterozygous in nonPAR region, then the genotype of that male sample is considered missing and will be imputed using mean genotype.
• If a male has genotype coded as ./x in nonPAR region, then the genotype of that male sample is considered missing and will be imputed using mean genotype.
• For details of methods analyzing chromosomeX, please refer to method.