Assessing multivariate gene-metabolome associations with rare variants using Bayesian reduced rank regression.

We considered a replication significant if the test was nominally significant ( P < 0.05) in both replication datasets, and the P -value for the pooled estimate was significant after correcting for the multiple tests.

also in 7,414 other papers

For example, when we investigated the results more carefully, we discovered SNPs in which the minor allele was present in few individuals, and such individuals could be almost perfectly identified by a seemingly random combination of phenotypes, leading to a large spurious canonical correlation test score (or, equivalently, a highly significant P -value).

also in 127,188 other papers