Machine Learning Consensus Scoring Improves Performance Across Targets in Structure-Based Virtual Screening.

In structure-based virtual screening, compound ranking through a consensus of scores from a variety of docking programs or scoring functions, rather than ranking by scores from a single program, provides better predictive performance and reduces target performance variability. Here we compare traditional consensus scoring methods with a novel, unsupervised gradient boosting approach. We also observed increased score variation among active ligands and developed a statistical mixture model consensus score based on combining score means and variances. To evaluate performance, we used the common performance metrics ROCAUC and EF1 on 21 benchmark targets from DUD-E. Traditional consensus methods, such as taking the mean of quantile normalized docking scores, outperformed individual docking methods and are more robust to target variation. The mixture model and gradient boosting provided further improvements over the traditional consensus methods. These methods are readily applicable to new targets in academic research and overcome the potentially poor performance of using a single docking method on a new target.

Structural Analysis of Multi-Helical RNAs by NMR-SAXS/WAXS: Application to the U4/U6 di-snRNA.

NMR and SAXS (small-angle X-ray scattering)/WAXS (wide-angle X-ray scattering) are highly complementary approaches for the analysis of RNA structure in solution. Here we describe an efficient NMR-SAXS/WAXS approach for structural investigation of multi-helical RNAs. We illustrate this approach by determining the overall fold of a 92-nt 3-helix junction from the U4/U6 di-snRNA. The U4/U6 di-snRNA is conserved in eukaryotes and is part of the U4/U6.U5 tri-snRNP, a large ribonucleoprotein complex that comprises a major subunit of the assembled spliceosome. Helical orientations can be determined by X-ray scattering data alone, but the addition of NMR RDC (residual dipolar coupling) restraints improves the structure models. RDCs were measured in two different external alignment media and also by magnetic susceptibility anisotropy. The resulting alignment tensors are collinear, which is a previously noted problem for nucleic acids. Including WAXS data in the calculations produces models with significantly better fits to the scattering data. In solution, the U4/U6 di-snRNA forms a 3-helix junction with a planar Y-shaped structure and has no detectable tertiary interactions. Single-molecule Förster resonance energy transfer data support the observed topology. A comparison with the recently determined cryo-electron microscopy structure of the U4/U6.U5 tri-snRNP illustrates how proteins scaffold the RNA and dramatically alter the geometry of the U4/U6 3-helix junction.

Water, shape recognition, salt bridges, and cation-pi interactions differentiate peptide recognition of the HIV rev-responsive element.

Recognition of the human immunodeficiency virus Rev-responsive element (RRE) RNA by the Rev protein is an essential step in the viral life cycle. Formation of the Rev-RRE complex signals nucleocytoplasmic export of unspliced and partially spliced viral RNA. Essential components of the complex have been localized to a minimal arginine-rich Rev peptide and stem IIB of RRE. In vitro selection studies have identified a synthetic peptide known as RSG 1.2 that binds with better specificity and affinity to RRE than the Rev peptide. NMR structures of both peptide-RNA complexes of Rev and RSG 1.2 bound to RRE stem IIB have been solved and reveal gross structural differences between the two bound complexes. Molecular dynamics simulations of the Rev and RSG 1.2 peptides in complex with RRE stem IIB have been simulated to better understand on an atomic level how two arginine-rich peptides of similar length recognize the same sequence of RNA with such different structural motifs. While the Rev peptide employs some base-specific hydrogen bonding for recognition of RRE, shape recognition, through contact with the sugar-phosphate backbone, and cation-pi interactions are also important. Molecular dynamics simulations suggest that RSG 1.2 binds more tightly to the RRE sequence than Rev by forming more base-specific contacts, using water to mediate peptide-RNA contacts, and is held in place by a strong salt bridge network spanning the major groove of the RNA.