Unidirectional rare variant gene-based tests
|
Collapsing methods
|
Combined Multivariate and Collapsing (CMC) |
Liu & Leal, PLoS Comp. Bio. 2008 |
EPACTS |
All rare variants collapsed into a single variant; individual dosage for the collapsed ‘variant’ is regressed against phenotype. |
Weighted and un-weighted sum methods
|
Variable threshold (VT) |
Price et al, AJHG. 2010 |
PLINK-Seq |
Sum of rare allele count in cases vs. controls; allele frequency threshold for inclusion is varied to maximize test statistic. |
Weighted Sum Statistic (FRQWGT) |
Madsen & Browning, PLoS Gen. 2009 |
PLINK-Seq |
Permutation-based test comparing inverse-frequency-weighted rare variant counts per individual in cases vs. controls. |
Weighted Sum Method (WILCOX-WSS) |
Madsen & Browning, PLoS Gen. 2009 |
EPACTS |
Wilcoxon Rank Sum test between phenotypes and inverse frequency-weighted rare variant scores. |
Kernel-Based Adaptive Cluster (KBAC) |
Liu & Leal, PLoS Gen. 2010 |
PLINK-Seq |
Variant weights are determined adaptively, and are based on observed effect sizes; individuals scored by weighted sum of allele counts. |
Summary case:control count methods
|
BURDEN method |
Purcell (PLINK-Seq) |
PLINK-Seq |
Permutation-based test comparing raw allele counts in cases vs. controls. |
UNIQ test |
Purcell (PLINK-Seq) |
PLINK-Seq |
Simple count of total case-unique rare alleles; permutations to assess significance. |
Bi-directional variance-component gene-based tests
|
C-ALPHA |
Neale et al, PLoS Gen. 2011 |
PLINK-Seq |
Detects deviation of observed case:control variant counts from expected binomial distribution. |
Sequence Kernel Association Test (SKAT) |
Wu et al, AJHG 2011 |
EPACTS |
Generalized form of C-ALPHA with variants weighted by allele frequency. |
Linear combination of unidirectional and variance-component tests
|
SKAT-O (‘Optimal’ SKAT) |
Lee et al, AJHG. 2012 |
EPACTS |
Adaptive linear combination of unidirectional burden test and variance-component SKAT test. |
Mixed Effects Score Test (MiST) |
Sun et al, Genetic Epi. 2013 |
Public R package |
Hierarchical regression model combining two independent test statistics which quantify variant effect sizes and ‘heterogeneity’. |