RNA-Puzzles Round V: Models from LCBio group
Description
This dataset comprises the three-dimensional RNA structure models submitted by the LCBio group for the RNA-Puzzles Round V blind prediction challenge. The collection includes computational predictions for seven distinct RNA targets: PZ30, PZ31, PZ32, PZ33, PZ34, PZ35, and PZ36. The models were generated using the LCBio group's pipeline, which combines consensus secondary structure prediction with coarse-grained 3D modeling and refinement. The LCBio approach focuses on: - Secondary Structure Prediction: A consensus secondary structure was derived using multiple methods (ViennaRNA, RNAStructure, ProbKnot, CentroidFold, ContraFold, IPknot) and homologous sequence information from Rfam and RNAcentral. - 3D Modeling: The SimRNA program was used for 3D structure modeling via replica-exchange Monte Carlo simulations, performed both with and without secondary structure restraints. - Refinement: The lowest-energy coarse-grained models were selected and refined using the QRNAs program to reconstruct all-atom details and mitigate modeling errors. - Homology Modeling: For specific targets like PZ30, homology modeling was employed based on identified structural homologs. Each model reflects the group's methodology in predicting tertiary structures from primary sequences, contributing to the community-wide effort to improve RNA structure prediction accuracy. These submissions were part of the RNA-Puzzles Round V assessment, a blind challenge where participants predict structures before the experimental results are released. This dataset provides a valuable resource for researchers interested in benchmarking RNA folding software, analyzing prediction performance, or studying the structural characteristics of the specific targets involved in this round of the challenge.
Files
Steps to reproduce
For all puzzles, homologous RNA sequences and 3D structures were identified using Rfam, RNAcentral, and RCSB PDB databases. Secondary structure predictions were performed using a consensus approach involving ViennaRNA, RNAStructure, ProbKnot, CentroidFold, ContraFold, and IPknot. The 3D structures were then modeled using SimRNA with replica-exchange Monte Carlo simulations, both with and without secondary structure restraints. The lowest-energy structures from the top clusters were selected and subsequently refined using the QRNAs program to reduce coarse-grained modeling errors. For specific cases like PZ30, homology modeling was also employed. Molecular dynamics simulation was conducted to refine the models, with the exception of PZ30.For detailed description of the protocol, please refer to the associated publication (see "LCBio group" section in Methods).
Institutions
- University of WarsawMazovia, Warsaw
Categories
Funders
- Polish Grid InfrastructureLesser Poland, WroclawGrant ID: PLG/2022/016043; PLG/2022/015327; PLG/2020/013424