**0**wrote:

Hi list,

My question is simple: **what questions are each of these two tests responding to?** (why do i get different results when testing for the same coefficient in the Wald vs LRT test?) The FC is the same, the p-values change because the LRT provides a p-value testing for the null hypothesis that the coefficients that are not in the reduced formula are equal to zero, the Wald test provide p-values testing for the null hypothesis that the coefficients of the full model are equal to zero. But how do you interpret this?

*What are the genes that are being shown as significant for each test?*

> design(ddsK)=~A*B*C > resRed = DESeq(ddsK, test = "LRT", reduced = formula(~ A*B)) > resRed = results(resRed, name = "C_C2_vs_C1", alpha = 0.05) > summary(resRed) out of 28842 with nonzero total read count adjusted p-value < 0.05 LFC > 0 (up) : 5936, 21% LFC < 0 (down) : 5894, 20% outliers [1] : 81, 0.28% low counts [2] : 3354, 12% (mean count < 1) [1] see 'cooksCutoff' argument of ?results [2] see 'independentFiltering' argument of ?results > design(ddsK)=~A*B*C > ddsK=DESeq(ddsK) > resB <- results(ddsK, name = "C_C2_vs_C1", alpha = 0.05) > summary(resB) out of 28842 with nonzero total read count adjusted p-value < 0.05 LFC > 0 (up) : 1142, 4% LFC < 0 (down) : 1088, 3.8% outliers [1] : 81, 0.28% low counts [2] : 5015, 17% (mean count < 3) [1] see 'cooksCutoff' argument of ?results [2] see 'independentFiltering' argument of ?results > resRed log2 fold change (MLE): C C2 vs C1 LRT p-value: '~ A * B * C' vs '~ A * B' DataFrame with 29085 rows and 6 columns baseMean log2FoldChange lfcSE stat pvalue padj <numeric> <numeric> <numeric> <numeric> <numeric> <numeric> comp100000_c0 5.394950 0.88129957 0.7296606 25.557624 3.885476e-05 1.842103e-04 comp100000_c1 15.236690 0.83653169 0.6352388 48.670559 6.839876e-10 8.012020e-09 comp100000_c3 7.684048 -0.02531354 0.9073333 23.174532 1.168546e-04 4.968079e-04 comp100002_c0 241.680116 1.27044149 0.3719448 52.028340 1.360758e-10 1.797857e-09 comp100003_c0 11.010333 1.44552180 0.6997581 8.617796 7.139597e-02 1.266642e-01 ... ... ... ... ... ... ... comp99991_c0 1643.4060 0.06606909 0.08068866 39.811182 4.735591e-08 4.018609e-07 comp99992_c0 651.1968 -0.08546246 0.19208340 2.133650 7.111930e-01 7.639641e-01 comp99993_c0 1957.2871 0.08239842 0.13402428 5.188685 2.684794e-01 3.605124e-01 comp99994_c0 1562.8709 -0.12381266 0.14475235 1.837816 7.655570e-01 8.095270e-01 comp99999_c0 170.9605 0.15237421 0.20194552 4.054691 3.986549e-01 4.888804e-01 > resB log2 fold change (MLE): C C2 vs C1 Wald test p-value: C C2 vs C1 DataFrame with 29085 rows and 6 columns baseMean log2FoldChange lfcSE stat pvalue padj <numeric> <numeric> <numeric> <numeric> <numeric> <numeric> comp100000_c0 5.394950 0.88129957 0.7296606 1.20782123 0.2271160328 0.47320324 comp100000_c1 15.236690 0.83653169 0.6352388 1.31687747 0.1878796988 0.42510643 comp100000_c3 7.684048 -0.02531354 0.9073333 -0.02789883 0.9777428379 0.98894584 comp100002_c0 241.680116 1.27044149 0.3719448 3.41567233 0.0006362476 0.01283631 comp100003_c0 11.010333 1.44552180 0.6997581 2.06574505 0.0388525593 0.18057827 ... ... ... ... ... ... ... comp99991_c0 1643.4060 0.06606909 0.08068866 0.8188151 0.4128919 0.6519404 comp99992_c0 651.1968 -0.08546246 0.19208340 -0.4449237 0.6563749 0.8259819 comp99993_c0 1957.2871 0.08239842 0.13402428 0.6148022 0.5386854 0.7499239 comp99994_c0 1562.8709 -0.12381266 0.14475235 -0.8553413 0.3923622 0.6347520 comp99999_c0 170.9605 0.15237421 0.20194552 0.7545313 0.4505303 0.6829395

**25k**• written 3.3 years ago by aziza •

**0**