For all possible absolute log2 fold changes (LFC), which genes have at least this fold change at a specified False Discovery Rate (FDR)?

limma_confects(fit, coef = NULL, fdr = 0.05, step = 0.001, trend = FALSE, full = FALSE)

fit | A limma MArrayLM object. |
---|---|

coef | Number or name of coefficient or contrast to test. |

fdr | False Discovery Rate to control for. |

step | Granularity of log2 fold changes to test. |

trend | Should |

full | Include some further statistics used to calculate confects in the output, and also include FDR-adjusted p-values that effect size is non-zero (note that this is against the spirit of the topconfects approach). |

See `nest_confects`

for details of how to interpret the result.

Results are presented in a table such that for any given LFC, if the reader
chooses the genes with `abs(confect)`

less than this they are assured
that this set of genes has at least this LFC (with the specified FDR). Once
this set of genes is selected, the confect values provide confidence bounds
with False Coverage-statement Rate at the same level as the FDR.

`fit`

should be produced using `lmFit`

. It is not necessary to use
`eBayes`

, this function calls `eBayes`

itself.

To test contrasts, this function can also be used with the result of
`contrasts.fit`

, but limma's handling of weights may be approximate (for
example if `voom`

has been used). For exact results for a contrast, use
`contrastToCoef`

to adjust the design matrix given to `lmFit`

.

#Prepare a data set library(NBPSeq) library(edgeR) library(limma) data(arab) dgelist <- DGEList(arab) dgelist <- calcNormFactors(dgelist) cpms <- cpm(dgelist, log=TRUE) # Retain genes with more than a geometric mean of 2 RPM # (about 5 reads per sample) cpms <- cpms[rowMeans(cpms) >= 1,] # Fit linear model for each gene treatment <- c(FALSE,FALSE,FALSE,TRUE,TRUE,TRUE) batch <- factor(c(1,2,3,1,2,3)) design <- model.matrix(~ treatment + batch) fit <- lmFit(cpms, design) # Calculate top confects # As voom has not been used, it is necessary to use trend=TRUE limma_confects(fit, "treatmentTRUE", trend=TRUE)#> $table #> rank index confect effect AveExpr name #> 1 16277 3.368 6.119901 4.562624 AT5G48430 #> 2 9681 2.955 4.737859 6.580063 AT3G46280 #> 3 5325 2.939 4.448903 6.074351 AT2G19190 #> 4 6738 2.728 4.279509 5.478560 AT2G39530 #> 5 6725 2.728 4.690481 4.339498 AT2G39380 #> 6 2741 2.728 4.967374 3.829884 AT1G51850 #> 7 11550 2.681 4.359977 9.145701 AT4G12500 #> 8 2738 2.654 3.975733 6.622807 AT1G51800 #> 9 10315 2.654 5.003027 3.347683 AT3G55150 #> 10 7112 2.654 4.864416 3.702749 AT2G44370 #> ... #> 1719 of 17681 non-zero log2 fold change at FDR 0.05 #> Prior df 10.6