AutoNOE Rosetta

This chapter describes the AutoNOE protocol [1]. AutoNOE is a combined automated NOE assignment and structure determination approach. In this approach, the NOE assignment algorithm makes use of chemical shift resonances to assign an initial set of NOESY cross-peaks. These NOE assignments are filtered based on a set of scores contributed by self-consistent restraint network, symmetric peaks, compatibility with chemical shifts and preliminary structures (if available). The intensities of cross-peaks for the filtered NOE assignments are calibrated to obtain distance restraints, which are further applied during structure calculation within RASREC. The structural models output by RASREC are used for next generation of batches to further strengthen assignments and structure refinement. This iterative process is carried out until converged structural models together with a modest set of long-range NOE restraints are obtained.

The underlying structure determination of AutoNOE is RASREC, which is described here in detail. The score terms used during the NOE assignment process and the architecture of the protocol are clearly explained in [1]. In addition, AutoNOE was tested on a benchmark set of 50 targets. The performance and accuracy of these tests can be found in [2].

AutoNOE can be combined with any chemical shift assignment algorithm to create a fully automated NMR structure determination pipeline. This was shown recently by combining 4D-CHAINS resonance assignment algorithm with AutoNOE on a benchmark set of four targets. The details of this work can be found in [3].

Preparation of Input Files

AutoNOE uses the same input files as the basic CS-Rosetta protocol and an additional set of unassigned NOESY peak lists.

Running AutoNOE CS-Rosetta

To prepare a AutoNOE simulation, perform the same setup steps as described before, but use setup_target -method autoNOE for your method-choice and provide the additional unassigned NOESY peak lists using the option –peaks peak_list1 peak_list2…. AutoNOE runs parallel structure calculations in batches using RASREC's work-load manager. Sub-directories 'batch_000001', 'batch_000002', 'batch_000003' are automatically generated when a new batch is started and are populated with all necessary input files, e.g., batch-dependent flags, broker setup, structural input information and NOE assignment data (restraints and peak assignment confidence scores). The calculated structures of a batch are generally stored in batch_nnnnn/decoys.out. Intermediate structures after completion of abinitio-stages (stage1, stage2, stage3 and stage4) are stored in decoys_stage1.out,..., decoys_stage4.out, accordingly.

RASREC's workload manager performs the same set of steps as it would without AutoNOE protocol. The key part of AutoNOE protocol is the generation of NOE restraint file and an additional set of best ranked structures in each batch of runs within RASREC for subsequent rounds of NOE assignment. To generate the NOE restraints, AutoNOE protocol first performs an initial set of assignments using chemical shift resonances and NOE cross peaks. From this set, it eliminates any diagonal peaks (NOE between protons of the same residue). The remaining assignments are scored first using given chemical shifts, distance compliance and agreement with models (termed decoy compliance) produced using RASREC. Subsequently, the assignment confidence is strengthened by looking for symmetric peaks and a self-consistent network of restraints. Following all these steps, NOE restraints are generated. Refer [1] for details about the parameters used in calculations and to learn how RASREC and AutoNOE communicate.

Restarting AutoNOE calculations

Similar to RASREC calculations, AutoNOE calculations can be restarted without the need of further interaction when stopped prematurely. See Restarting RASREC calculations for details.

Post-Processing of AutoNOE structure calculations

AutoNOE simulation is finished if directory fullatom_pool_stage8 exists in all the run directories of the cst_XX folders. To pick the best constraint weight run, execute the autoNOE_select_final_run script. This script picks the best weight by analyzing the energies of the final models, fraction of residues converged and the quality of NOEs assigned. The final decoys within the best run will be in the silent file under cst_XX /run/fullatom_pool/decoys.out. Additionally, the fullatom_pool directory consists of _auto_analysis folder, which has the 10 lowest-energy structures along with their statistics.

For more information see the respective Tutorials and the Application Reference.


sampling protocol: