Rotamer Libraries for RNA-Binding Proteins (RBPs) in Bound and Unbound States

Overview This repository contains a comprehensive set of rotamer libraries for RNA-Binding Proteins (RBPs), specifically curated to analyze side-chain conformational preferences in the context of protein-RNA interactions. The libraries are stratified based on: Backbone dependence: Backbone-Dependent (BBD) and Backbone-Independent (BBI) libraries. Functional site: Residues at the RNA-binding interface (I) vs. non-interface (N) sites. Functional state: Protein structures in the bound (B) and unbound (U) conformations. These libraries are essential for computational structural biology, particularly in improving the accuracy of protein-RNA docking, side-chain packing prediction, and molecular modeling tasks. Key Features State-Specific Data: Separate statistics for bound and unbound protein conformations. Spatially Resolved: Distinct data for interface and non-interface residues. Statistically Robust: Probabilities and side-chain dihedral angles (χ) are provided as weighted circular means and standard deviations. Dataset Structure and File Descriptions The dataset is organized into two main catagories (with "bbd" for backbone dependent and "bbi" for backbone independent rotamers): RBPs_rotamer_libraries/ ├── 1_BBD_rotamer_lib/ # Backbone-Dependent Library │ ├── RBPs_I_bbd_B_rotamer_lib.csv # Interface, Bound │ ├── RBPs_I_bbd_U_rotamer_lib.csv # Interface, Unbound │ ├── RBPs_N_bbd_B_rotamer_lib.csv # Non-Interface, Bound │ ├── RBPs_N_bbd_U_rotamer_lib.csv # Non-Interface, Unbound │ ├── RBPs_bbd_B_rotamer_lib.csv # Overall, Bound │ └── RBPs_bbd_U_rotamer_lib.csv # Overall, Unbound └── 2_BBI_rotamer_lib/ # Backbone-Independent Library ├── RBPs_I_bbi_B_rotamer_lib.csv # Interface, Bound ├── RBPs_I_bbi_U_rotamer_lib.csv # Interface, Unbound ├── RBPs_N_bbi_B_rotamer_lib.csv # Non-Interface, Bound ├── RBPs_N_bbi_U_rotamer_lib.csv # Non-Interface, Unbound ├── RBPs_bbi_B_rotamer_lib.csv # Overall, Bound └── RBPs_bbi_U_rotamer_lib.csv # Overall, Unbound Backbone-Dependent (BBD) Library Columns Each row corresponds to a specific (φ, ψ) bin. Column Name Description AA Amino acid type (3-letter code). PHI Backbone φ torsion angle (center of bin, in degrees). PSI Backbone ψ torsion angle (center of bin, in degrees). _Count Number of data points in this specific (φ, ψ) bin. Count Total number of data points in the (φ, ψ) neighborhood used for smoothing. Prob Probability of the rotamer in this bin (range 0.0-1.0). B/U_CHI[1-4] Bound/Unbound χₙ weighted circular mean (degrees). B/U_CHI[1-4]Sig Bound/Unbound χₙ weighted circular standard deviation (degrees). Backbone-Independent (BBI) Library Columns Each row provides global statistics for an amino acid type. Column Name Description AA Amino acid type (3-letter code). Count Total number of data points for this amino acid. Prob Normalized probability of occurrence (range 0.0-1.0). B/U_CHI[1-4] Bound/Unbound χₙ weighted circular mean (degrees). B/U_CHI[1-4]Sig Bound/Unbound χₙ weighted circular standard deviation (degrees). Generation Methodology These libraries were constructed from a high-quality, non-redundant set of protein-RNA complexes and their corresponding unbound structures. The methodology is based on established practices in rotamer library development: Data Curation: Structures were sourced from the Protein-RNA Docking Benchmark v3.0. Interface Definition: Residues were classified as interface (I) or non-interface (N) based on solvent accessibility changes and distance criteria to RNA upon binding. Dihedral Calculation: Side-chain χ angles were calculated and binned. Statistical Smoothing: Probabilities, circular means, and standard deviations were calculated using adaptive kernel density estimates and methods from circular statistics. Notes All angular measurements are in degrees. Probability values (Prob) range from 0.0 (never observed) to 1.0 (always observed). Standard deviations are calculated using circular statistics methods suitable for angular data. The "Overall" libraries (RBPs_bbd_*, RBPs_bbi_*) contain data from all residues, regardless of their interface classification.