Highlights d A deep learning model for AD prediction was derived from a large set of synthetic ADs d The predictor (ADpred) identifies sequence features important for acidic AD function d AD sequence features explain the basis for the fuzzy binding mechanism of acidic ADs d Acidic ADs are enriched in yeast but not in Drosophila or human transcription factors

Deletions within genes coding for subunits of the transcription coactivator SAGA caused strong genome-wide defects in transcription and SAGA-mediated chromatin modifications. In contrast, rapid SAGA depletion produced only modest transcription defects at 13% of protein-coding genes – genes that are generally more sensitive to rapid TFIID depletion. However, transcription of these ‘coactivator-redundant’ genes is strongly affected by rapid depletion of both factors, showing the overlapping functions of TFIID and SAGA at this gene set. We suggest that this overlapping function is linked to TBP-DNA recruitment. The remaining 87% of expressed genes that we term ‘TFIID-dependent’ are highly sensitive to rapid TFIID depletion and insensitive to rapid SAGA depletion. Genome-wide mapping of SAGA and TFIID found binding of both factors at many genes independent of gene class. Promoter analysis suggests that the distinction between the gene classes is due to multiple components rather than any single regulatory factor or promoter sequence motif.

No abstract

Protein-protein interactions (PPI) in nature are conveyed by a multitude of binding modes involving various surfaces, secondary structure elements and intermolecular interactions. This diversity results in PPI binding affinities that span more than nine orders of magnitude. Several early studies attempted to correlate PPI binding affinities to various structure-derived features with limited success. The growing number of high-resolution structures, the appearance of more precise methods for measuring binding affinities and the development of new computational algorithms enable more thorough investigations in this direction. Here, we use a large dataset of PPI structures with the documented binding affinities to calculate a number of structure-based features that could potentially define binding energetics. We explore how well each calculated biophysical feature alone correlates with binding affinity and determine the features that could be used to distinguish between high-, medium- and low- affinity PPIs. Furthermore, we test how various combinations of features could be applied to predict binding affinity and observe a slow improvement in correlation as more features are incorporated into the equation. In addition, we observe a considerable improvement in predictions if we exclude from our analysis low-resolution and NMR structures, revealing the importance of capturing exact intermolecular interactions in our calculations. Our analysis should facilitate prediction of new interactions on the genome scale, better characterization of signaling networks and design of novel binding partners for various target proteins.

No abstract