AFFINITY DENSITY: A NOVEL GENOMIC APPROACH TO THE IDENTIFICATION OF TRANSCRIPTION FACTOR REGULATORY TARGETS
We developed a new (at the time) method for application of signal processing techniques to transcription factor specificity on the genome. We found that the approach allowed us to identify known regulatory targets of a transcription factor involved in animal heart development. We also showed the power to predict targets in an unbiased gene knockdown screen.
Abstract
METHODS:
A new method was developed for identifying novel transcription factor regulatory targets based on calculating Local Affinity Density. Techniques from the signal-processing field were used, in particular the Hann digital filter, to calculate the relative binding affinity of different regions based on previously published in vitro binding data. To illustrate this approach, the complete genomes of Drosophila melanogaster and D.pseudoobscura were analyzed for binding sites of the homeodomain proteinc Tinman, an essential heart development gene in both Drosophila and Mouse. The significant binding regions were identified relative to genomic background and assigned to putative target genes. Valid candidates common to both species of Drosophila were selected as a test of conservation.
RESULTS:
The new method was more sensitive than cluster searches for conserved binding motifs with respect to positive identification of known Tinman targets. Our Local Affinity Density method also identified a significantly greater proportion of Tinman-coexpressed genes than equivalent, optimized cluster searching. In addition, this new method predicted a significantly greater than expected number of genes with previously published RNAi phenotypes in the heart.
AVAILABILITY:
Algorithms were implemented in Python, LISP, R and maxima, using MySQL to access locally mirrored sequence data from Ensembl (D.melanogaster release 4.3) and flybase (D.pseudoobscura). All code is licensed under GPL and freely available at http://www.ohsu.edu/cellbio/dev_biol_prog/affinitydensity/.
RELEVANCE TO OC
This study helped lay a foundation for our understanding of transcription factor binding to DNA, and forms the basis of our understanding of variant effects and other models of gene regulation in ovarian cancer.
PUBLICATION LINK