Kriging-style LD-consistency outlier QC

Flags variants whose observed z-score is inconsistent with the value predicted from its LD neighbours, using susieR's kriging diagnostic. susieR::kriging_rss() computes the leave-one-out conditional distribution of each z_i given the rest (with the LD-mismatch scale s defaulting to susieR::estimate_s_rss()) and a per-variant logLR for the allele-switch hypothesis. This helper reuses susieR's own allele-switch rule — logLR > logLRThreshold & abs(z) > zThreshold (the same logLR > 2 & |z| > 2 used in susie_rss_utils) — to flag variants whose sign should be flipped. RSS-only helper, opt-in via alleleFlipKriging; never wired into alleleQc() / matchRefPanel(). Requires a susieR that provides kriging_rss() and estimate_s_rss().

Usage

krigingOutlierQc(
  zScore,
  R,
  n,
  variantIds = NULL,
  zThreshold = 2,
  logLRThreshold = 2,
  methodArgs = KrigingOptions()
)

Arguments

zScore

Numeric vector of harmonized z-scores.

R

Square LD correlation matrix aligned to zScore.

n

Sample size, forwarded to susieR::kriging_rss() (whose default s is susieR::estimate_s_rss()).

variantIds

Optional variant IDs for the diagnostics table.

zThreshold

Absolute-z cutoff for the allele-switch rule (default 2, matching susieR).

logLRThreshold

Log-likelihood-ratio cutoff for the allele-switch rule (default 2, matching susieR).

methodArgs

Extra arguments for susieR::kriging_rss(), built with KrigingOptions – r_tol and s.

Value

A list with flip (logical vector; TRUE = allele switch, z-score should be sign-flipped) and diagnostics (data frame of per-variant z, condmean, z_std_diff, logLR, and the flipped flag).

Examples

data(eqtlRegionExample)
X <- eqtlRegionExample$X[, 1:20]
R <- cor(X)
krigingOutlierQc(
  zScore = rnorm(20), R = R, n = 415, variantIds = colnames(X))
#> $flip
#>  [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
#> [13] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
#> 
#> $diagnostics
#> # A tibble: 20 × 6
#>    variant_id               z condmean z_std_diff    logLR flipped
#>    <chr>                <dbl>    <dbl>      <dbl>    <dbl> <lgl>  
#>  1 chr22:32119788:T:C -0.291    0.173     -0.471   0.0897  FALSE  
#>  2 chr22:32119867:T:G -0.114    0.159     -0.277   0.0323  FALSE  
#>  3 chr22:32119961:T:G -2.55    -0.119     -2.44   -0.531   FALSE  
#>  4 chr22:32120053:T:C  0.653   -0.128      0.792   0.148   FALSE  
#>  5 chr22:32120593:A:G -1.62     0.0604    -1.70    0.174   FALSE  
#>  6 chr22:32120636:T:C -1.47     0.147     -1.62    0.378   FALSE  
#>  7 chr22:32120932:G:A -0.430    0.0934    -0.527   0.0704  FALSE  
#>  8 chr22:32120975:A:G  0.964    0.0462     0.925  -0.0782  FALSE  
#>  9 chr22:32121290:C:T  0.0767   0.299     -0.227  -0.0413  FALSE  
#> 10 chr22:32121498:A:C  1.30     0.214      1.11   -0.500   FALSE  
#> 11 chr22:32121635:T:G -0.0892  -0.0599    -0.0298 -0.00949 FALSE  
#> 12 chr22:32121702:A:G -1.09    -0.145     -0.952  -0.277   FALSE  
#> 13 chr22:32121936:G:A  0.249    0.162      0.0883 -0.0703  FALSE  
#> 14 chr22:32122351:A:G  1.39     0.200      1.21   -0.500   FALSE  
#> 15 chr22:32122676:A:G -0.0794  -0.0607    -0.0191 -0.00857 FALSE  
#> 16 chr22:32123055:A:G  1.25     0.0193     1.24   -0.0425  FALSE  
#> 17 chr22:32123164:T:C -0.357    0.282     -0.650   0.180   FALSE  
#> 18 chr22:32123383:T:C -0.711    0.340     -1.07    0.434   FALSE  
#> 19 chr22:32123461:T:C -1.24    -0.0481    -1.20   -0.103   FALSE  
#> 20 chr22:32123806:C:T  1.24     0.202      1.06   -0.452   FALSE  
#>