Fast probabilistic whitening transformation for ultra-high dimensional genetic data.

In schizophrenia GWAS z-statistic imputation, probabilistic whitening yields the lowest mean square error while running up to an order of magnitude faster than other whitening methods.

• Why it matters: Correlation from linkage disequilibrium between nearby variants breaks independence assumptions and complicates downstream analyses that rely on regression summary statistics.
• What they did: The study proposes a probabilistic model for data whitening, with parameter estimates and efficient algorithms, and applies it to impute z-statistics using LD reference panels.
• The result: The approach improves statistical and computational performance and is used to identify tandem repeats that explain genetic regulatory signals for disease-relevant genes.