Abstract
First, we will discuss a new method, Virtual ChiP-seq which predicts binding of individual transcription factors in new cell types using an artificial neural network that integrates ChiP-seq results from other cell types and chromatin accessibility data in the new cell type. Virtual ChiPseq also uses learned associations between gene expression and transcription factor binding at specific genomic regions. This approach outperforms methods that use transcription factor sequence preferences in the form of position weight matrices, predicting binding for 39 transcription factors (area under receiver operating characteristic curve >0.97; area under precision-recall curve >0.3).
Second, we will discuss a new method to discover transcription factor motifs and identify transcription factor binding sites in DNA with covalent modifications such as methylation. Just as transcription factors distinguish one standard nucleobase from another, they also distinguish unmodified and modified bases. To represent the modified bases in a sequence, we replace cytosine (C) with symbols for 5-methylcytosine (5mC), and 5-hydroxylmethylcytosine (5hmC). Similarly, we adapted the well-established position weight matrix model of transcription factor binding affinity to an expanded alphabet. We created an expanded-alphabet genome sequence using genome-wide maps of 5mC, and 5hmC in mouse naive T cells. Using this sequence and expanded-alphabet position weight matrixes, we reproduced various known methylation binding preferences, including the preference of ZFP57 and C/EBPI3 for methylated motifs and the preference of c-Myc for unmethylated motifs. Using these known binding preferences to tune model parameters enables discovery of novel modified motifs.
Hosted by