Contents
compute$genome$cumulativeWeight
Cumulative PRS weight along the genome
Sums the absolute harmonized effect weight |beta| a model library places at each
variant -- a weighted alternative to the plain reuse count.
Usage
compute$genome$cumulativeWeight(models, min.models = 1)Arguments
| Argument | Description |
|---|---|
models | A PGSLibrary or a single PGS. Anything else aborts, as does a set mixing homogeneity strata. |
min.models | Numeric scalar, default 1. Minimum number of models a variant must appear in to be kept. Must be at least 1, and the call aborts when no variant clears it. |
Value
A PolyGeniusGenomeSignal with statistic = "cumulative.weight",
resolution = "variant", bin.reduce = "sum" and frame = "absolute". $results has
one row per retained variant, with chr, position, nea, ea, value
(sum(|beta|)) and n.models; $diagnostics holds n.ambiguous, the
strand-ambiguous model rows the harmonizer dropped.
Details
Cross-model scale. Absolute effect weights are not comparable across models built
from different traits, phenotype scales or algorithms (log-odds versus standardized
versus raw units; marginal C+T weights versus shrunken LDpred2/lassosum2 posteriors) --
see visualize$genome$effects. The sum is therefore dominated
by models carrying large-scale betas and, per display bin, by variant-dense regions; it
is a within-comparable-library heuristic, not a cross-trait weight. Read it alongside the
retained n.models to separate weight from count.
Because the statistic is a cross-model magnitude sum, models must be one homogeneity
stratum: the call aborts when two models carry different recorded GWAS sources,
generation rules or effect scales. A component no model records never trips the check.
Examples
sig <- compute$genome$cumulativeWeight(data$library, min.models = 2)
visualize$genome$cumulativeWeight(sig)