correlateGenes {scran}R Documentation

Per-gene correlation statistics

Description

Compute per-gene correlation statistics by combining results from gene pair correlations.

Usage

correlateGenes(stats)

Arguments

stats

A DataFrame returned by correlatePairs.

Details

For each gene, all of its pairs are identified, and the corresponding p-values are combined using Simes' method. This tests whether the gene is involved in significant correlations to any other gene. Per-gene statistics are useful for identifying correlated genes without regard to what they are correlated with (e.g., during feature selection).

Value

A DataFrame with one row per unique gene in stats and containing the fields:

gene:

A field of the same type as stats$gene1, specifying the identity of the gene corresponding to each row.

rho:

Numeric, the correlation with the largest magnitude across all gene pairs involving the corresponding gene.

p.value:

Numeric, the Simes p-value for this gene.

FDR:

Numeric, the adjusted p.value across all rows.

limited:

Logical, indicates whether the combined p-value is at its lower bound.

Author(s)

Aaron Lun

References

Simes RJ (1986). An improved Bonferroni procedure for multiple tests of significance. Biometrika 73:751-754.

See Also

correlatePairs to compute stats.

Examples

set.seed(0)
ncells <- 100
null.dist <- correlateNull(ncells, iters=100000)
exprs <- matrix(rpois(ncells*100, lambda=10), ncol=ncells)
out <- correlatePairs(exprs, null.dist=null.dist)

g.out <- correlateGenes(out)
head(g.out)

[Package scran version 1.12.1 Index]