Mr BUMP Automated Molecular Replacement Ronan Keegan

Mr. BUMP – Automated Molecular Replacement Ronan Keegan, Martyn Winn CCP 4, Daresbury Laboratory 28 th March 2007

Aim of Mr. BUMP • Automated framework for Molecular Replacement • Particular emphasis on generating variety of search models • • Wraps Phaser, Molrep and Acorn Uses a variety of helper applications (eg Chainsaw) and bioinformatics tools (eg FASTA, Mafft) • Uses on-line databases (eg PDB, Scop) • • In favourable cases, gives “one-button” solution In unfavourable cases, suggests likely search models for manual investigation 28 th March 2007

Pipeline Target MTZ & Sequence Target ` Details Template ` Search Check scores and exit or select the next model Model ` Preparation Molecular Replacement ` & Refinement 28 th March 2007

Template Search • Sequence based search (FASTA) • Secondary structure based search (SSM) • Domain search (SCOP) • Identification of possible multimers (PQS & PISA) • Users can also enter their own templates by ID or from locally held files. 28 th March 2007

Model Preparation • Search models can be prepared for MR in several ways – Chainsaw – non-conserved residues are pruned (sequence provided) – Molrep – pruning of non-conserved side-chains (internal sequence alignment) – Polyalanine – all side chain atoms are pruned beyond the CB atom – PDBclip – models are not modified • An ensemble of the best models is also created for Phaser 28 th March 2007

Molecular Replacement & Refinement • • • For each search model, MR is done with Molrep or Phaser or both. MR programs run mostly with defaults Mr. BUMP provides LABIN columns, MW of target, sequence identity of search model, number of copies to search for, number of clashes tolerated Allow Molrep / Phaser to set resolution limits and weights After MR, models are passed to Refmac for restrained refinement final Rfree < 0. 35 or final Rfree < 0. 5 and dropped by 20% “success” final Rfree < 0. 48 or final Rfree < 0. 52 and dropped by 5% “marginal” “failure” otherwise 28 th March 2007

Mr. BUMP on clusters • Mr. BUMP is usually run on a desktop • However, Mr. BUMP can take advantage of a compute cluster to farm out the Molecular Replacement jobs. • Currently Sun Grid Engine enabled clusters are supported but support will be added for other types of queuing system (e. g. LSF, Condor) if there is enough demand. • Job control: All nodes terminate when one finds a solution 28 th March 2007

Mr. BUMP Output • • • Currently produces a long log file listing search results, model preparation steps, summaries from each MR and refinement job and relevant references for programs used. Not ideal, there’s a lot of information to trawl through. Summary of results now provided at the end of log file. Future versions will provide results in marked-up web page format for more clarity. 28 th March 2007

Mr. BUMP pre-release • • • Beta version first released in Jan’ 06 (current version is 0. 3. 3) Currently supported on Linux and Mac OSX, Windows version will be available when included in suite. Will be included in next release of CCP 4 (version 6. 1) Mr. BUMP paper to be published in Acta Cryst. D in April ‘ 07 First citations in Obiero et al. , Acta Cryst. (2006). F 62, 757 -760; El Omari et al. , Acta Cryst. (2006). F 62, 949 -953 28 th March 2007

New features • Run Acorn after refinement for phase improvement (high resolution data) • Support for searching in enantiomorphic spacegroups. • Users can now specify template models by PDB ID or add local PDB files. • “Generate models only” option. • XML Output. • Additional multiple alignment programs supported – Tcoffee and Probcons. 28 th March 2007

New feartures: Smartie • Peter Briggs’ “Smart” CCP 4 log file browser. Robust method for reading and extracting information from log files. • Parses log files and picks out tables, header information and various other pieces of information. • Very useful when generating XML output from a automation wrapper script such as Mr. BUMP. 28 th March 2007

Future versions • Improvements to multimeric search models (using PISA) • Supplement multiple alignment with additional sequences and/or structural information • Model completion and/or re-building • Target complexes. 28 th March 2007

Conclusions • Test cases and the examples demonstrated the utility of trying a range of search models, a protocol that can only be attempted adequately by automation. • Mr. BUMP is not meant to compete with careful analysis of the data and model by an experienced crystallographer. However, it may succeed in difficult cases by finding a combination of models and protocols that would not otherwise have been tried. • In more straight forward cases the advantage is simply one of convenience. • CCP 4 web server log files indicate that it’s being used at least 34 times a day. 28 th March 2007

Some thoughts about automation • Mr. BUMP started out as a project to develop a one-button solution for MR (e-HTPX). It has evolved to be more interactive if required whilst retaining the one-button functionality. • Users like to have the donkey work done automatically but prefer controlled/transparent automation with the ability to both stop and start the program as they please at given points in the pipeline. • Also like to have a large degree of control of the input parameters. 28 th March 2007