Narlikar et al. 2010 employed LASSO regularization to identify heart-specific enhancers by using TF binding specificities from PSSMs as features [ 81], our approach in contrast used non-TF chromatin-associated features from ChIP-Seq data to identify enhancers.
Comparison with phage display and other experimental data suggests that the peptide extension approach recapitulates naturally occurring peptide binding specificity better than fixed backbone design, and that it should be useful for predicting peptide binding specificities from crystal structures.
The ability to accurately predict domain binding specificities from primary sequence in general, and for the PDZ domain in particular, would provide yet another step in that direction.
We have presented a new approach for learning regression-based models of protein DNA binding specificity from quantitative TF binding data, using SVR with feature selection.
This is again consistent with our previous hypothesis that pioneer factors derive their DNA binding specificity from multiple dimeric binding modes.
As for other transcription factors, attempts to model the binding specificity from few in vitro-assayed binding sites was met with variable success [ 27, 28].
Substitution of I1e-141 by Tyr-141 in the rat β-1 sGC changed its ligand binding specificity from NO to O2 [ 30].
It has been shown that the introduction of a polar Tyr residue into the non polar distal pocket of the H-NOB domain is sufficient to change its binding specificity from NO to O2.
The 1957 and 1968 pandemic H2 and H3 viruses were able to switch their binding specificity from SAα-2,3Gal to SAα-2,6Gal via two amino acid changes of Q226L and G228S (H3 numbering) within HA.
In conclusion, our regression-based approach for learning complex models of TF DNA binding specificity from custom PBM data can be easily extended and improved, and we anticipate that the proposed regression models will help explain, at least in part, how paralogous TFs with highly similar PWMs are able to interact with distinct genomic targets.
