Bujnicki lab - FiltRest3D
  • RNA has recently emerged as an attractive target for new drug development. Our team is developing new methods to study the interactions between RNA and ligands. Recently, we have developed a new machine learning method called AnnapuRNA to predict how small chemical molecules interact with structured RNA molecules. Research published in PLoS Comput Biol. 2021 Feb 1;17(2):e1008309. doi: 10.1371/journal.pcbi.1008309. Read More
  • 1

About Laboratory Of Bioinformatics And Protein Engineering

Our group is involved in theoretical and experimental research on nucleic acids and proteins. The current focus is on RNA sequence-structure-function relationships (in particular 3D modeling), RNA-protein complexes, and enzymes acting on RNA.
 
We study the rules that govern the sequence-structure-function relationships in proteins and nucleic acids and use the acquired knowledge to predict structures and functions for uncharacterized gene products, to alter the known structures and functions of proteins and RNAs and to engineer molecules with new properties.
 
Our key strength is in the integration of various types of theoretical and experimental analyses. We develop and use computer programs for modeling of protein three-dimensional structures based on heterogenous, low-resolution, noisy and ambivalent experimental data. We are also involved in genome-scale phylogenetic analyses, with the focus on identification of proteins that belong to particular families. Subsequently, we characterize experimentally the function of the most interesting new genes/proteins identified by bioinformatics. We also use theoretical predictions to guide protein engineering, using rational and random approaches. Our ultimate goal is to identify complete sets of enzymes involved in particular metabolic pathways (e.g. RNA modification, DNA repair) and to design proteins with new properties, in particular enzymes with new useful functions, which have not been observed in the nature.
 
We are well-equipped with respect to both theoretical and experimental analyses. Our lab offers excellent environment for training of young researchers in both bioinformatics and molecular biology/biochemistry of protein-nucleic acid interactions.


More Good Science

FiltRest3D - A Standalone Program

Michał J. Gajda, Marta Kaczor, Irina Tuszynska, Anastasia Yu. Bakulina, and Janusz M. Bujnicki
 
 
What is FiltRest3D?
 
 
Automatic methods for protein structure prediction (fold-recognition, de novo folding, and docking programs) produce large sets of alternative models. These large model sets often include many native-like structures, which are scored as high as false positives. Such native-like models can be more easily identified based on data from experimental analyses used as structural restraints (e.g. identification of nearby residues by crosslinking, chemical modification, site-directed mutagenesis, deuterium exchange coupled with mass spectrometry etc.). We present a simple server for scoring and ranking of models according to their agreement with user-defined restraints.
 
 
How to use it?
 
 
Program may be used through a web server, or downloaded and installed locally on any Linux* system, that is standard in most bioinformatics labs.
Source code is licensed on General Public License. Downloadable archive contains license, program code in Python, examples in example/ directory, and helper scripts in utils/. Minimum requirements is Python version >= 2.3, and BioPython library version >=1.41. Some functionality of the program my require installation of downloadable bioinformatics software, like Stride and DSSP - details are provided in installation instructions.
 
* Software should work correctly on any Unix system, including Mac OS-X, but authors didn't have an opportunity to test it on any other platform.
 
Examples of methods for model-building with the use of spatial restraints 

 
MolProbity is a web server recommended as a complementary resource for testing models for the presence of high-resolution features.
 
Example: discrimination of native-like complexes from low-resolutions docking decoys of pseudouridine synthase TruA from Thermus thermophilus.


The example file set of low-resolution docking decoys was obtained by a docking with GRAMM {Vakser 1995} of a TruA enzyme structure (PDB code 1VS3, apo form) to its tRNA substrate (PDB code 2V0G, a complex with an unrelated protein). As a reference we used the native structure of this complex (PBD code 2NR0)(Figure 1). We produced 30000 decoys with the grid step = 3.5 , and repulsion parameter = 20. The grid-step radius was used as a projection of an atom. The systematic search through the rotational coordinates was performed every 10 degree.

For discrimination of native-like complexes with FILTREST3D we used five distance restraints to prepare restraints file. First, we chose protein residues R50 and N52 known to be involved in catalysis of isomerization of U39 in tRNA and introduced two specific amino acid-nucleotide restraints. Second, we identified additional putative RNA-binding residues (R23, H119, R162.) that were both predicted as RNA-binding by the PPRINT webserver (Figure 2A) and were located in regions of positive electrostatic potential (Figure 2B), as calculated with the APBS tool with the PyMol program {DeLano 2002} for performing electrostatic calculations. For these three residues we defined a general restraint for interactions with any nucleotide of the whole tRNA molecule.


Then filtrest program was run:


Filtrest –r restr.txt –d list.txt– o filtrest.out


The filtering took 500 minutes with the standalone version of the program running on a Linux workstation with 3,06 GHz processor. In the output file 2 decoys satisfies restrains completely. They exhibit the RMSD to the native structure of 22.36 Å and 26.56 Å (Figure 3). They are similar to each other and can be considered native-like.


Figure 1: The native structure of pseudouridine synthase TruA in complex with leucyl tRNA, which was used as a reference.

 

 

Figure 2: The analysis of the protein surface of TruA. A – Regions, which according to PPRINT server could interact with RNA are colored yellow. B – Electrostatic map of the protein surface. The positive charged regions are blue, while negative charged regions are red.

Picture A                                                                                                       Picture B

 

Figure 3: A,B – Each of two best scored decoys, which were found by Filtrest3D superimposed on the reference structure of complex. Only proteins are superimposed. Reference structure is colored in blue tones, while decoys are colored in red tones. All atoms, which were used to make restraints have VDW representation. Cα of marked amino acids are colored white, while O3’ atoms of nucleic acids are yellow. C – Two best scored decoys. Proteins are superimposed.

Picture A

Picture B

 

Picture C


Restraints file syntax
 
A detailed manual for restraints file syntax is here. Online help is available for both web server and command line interface.