We would like to hypothesize how each TF contributes to accessibility changes while controlling for others. Then we can use a regression model to test the change in accessibility against whether motifs are present or not. If knowing the expression of TFs, we can better constrain the model such that consistent effects between TF expression and chromatin changes will be captured.
$y_1 = \beta[\text{TF}]_1\text{motif}_1$
$y_0 = \beta[\text{TF}]_0\text{motif}_1$
$\Delta y = \beta ([\text{TF}]_1 - [\text{TF}]_0)\text{motif}_1$
Compared with motif enrichment, this analysis specifically tests the effect of individual motif on changes in accessibility, which is driven by the binding of TFs. Between conditions, motifs are present or not stays the same due to same genetic sequence. However, the expression of TFs varies. Motif enrichment only tells us how each motif is found to be more present in the peaks compared to the background. The even naive way is to check the motif presence in differentially accessible peaks.
We would like to hypothesize how each TF contributes to accessibility changes while controlling for others. Then we can use a regression model to test the change in accessibility against whether motifs are present or not. If knowing the expression of TFs, we can better constrain the model such that consistent effects between TF expression and chromatin changes will be captured.
$y_1 = \beta[\text{TF}]_1\text{motif}_1$
$y_0 = \beta[\text{TF}]_0\text{motif}_1$
$\Delta y = \beta ([\text{TF}]_1 - [\text{TF}]_0)\text{motif}_1$
Compared with motif enrichment, this analysis specifically tests the effect of individual motif on changes in accessibility, which is driven by the binding of TFs. Between conditions, motifs are present or not stays the same due to same genetic sequence. However, the expression of TFs varies. Motif enrichment only tells us how each motif is found to be more present in the peaks compared to the background. The even naive way is to check the motif presence in differentially accessible peaks.