Automatic NOESY assignment Module (noesy_assign)

Author

Oliver Lange

last updated 5/31/11

Overview

In your code (application/protocol-level) you will probably only include the "NoesyModule" which exports the most important high-level funktions...

• NoesyModule – (top-level class) use one of these in your module/application to use the automatic NOESY assignment....

The NoesyModule class will read the input data: (assigned resonances and peak lists), assign them based on provided models or model-free and generate constraints.

The most important classes are:

• PeakFileFormat — handles input/output of CrossPeakLists (peak-lists)
• CrossPeakList — a peaklist ( peakID – resonances – assignments )
• ResonanceList — list of all resonances ( atomID - chemical shift value )
• ConstraintSet — holds all constraints generated

ResonanceList

• provides input/output of files with chemical shift assignment (.prot)
• each resonance has a unique identifier (ID / integer)
• each resonance combines a single resonance frequency, a tolerance with an atom (NamedAtomID)
• the list object also has the sequence information of protein

– resonance accessors: by ID, by resid (list) – sequence accesors: full sequence (string), at single position (AA)

a Resonance

• ID (unsigned integer)
• frequency + tolerance
• NamedAtomID

Assignment

the top-level procedure is found in NoesyModule.impl.hh: NoesyModule::assign()

• find_assignments — initial assignments by chemical shift matching
• delete_diagonal_peaks — delete peaks from list that have at least 1 assignment that is diagonal
• update assignment scores: update_chemicalshiftscore — compute score for chemical shift match update_symmetry_score update_upperdistance_score update_decoy_compatibility_score
• network_analysis –> yields network-score
• calibrate – peak-intensities -> distance bound
• peak-elimination —> heuristics to remove assignments that might be spurious

Data Structure

• CrossPeakList
• CrossPeak, CrossPeak3D, CrossPeak4D
• Spin
• PeakAssignment
• CrossPeakInfo

The CrossPeakList is simpy the list of all CrossPeaks

A CrossPeak is a 2D peak that is extended two 3D and 4D by inheritance The CrossPeak has two proton "Spin" and for 3D and 4D it has one or two label "Spin", respectively. Moreover the CrossPeak has a volume (Real), an peak_id (integer), and for the two protons with/without label it has a CrossPeakInfo field each: info1, info2. The CrossPeak further provides fields to store some computed values: -assignments -cumulative_peak_volume ( ess. the sum of all assignment scores used for normalization of these ) -distance_bound ( computed after peak-calibration ) -eliminated ( bool – various heuristics can cause elimination ) -elimination-reason

The assignments of a CrossPeak are stored in a distributed fashion: each Spin ( which a CrossPeak can have 2,3, or 4 ) has a list of possible assignments (i.e., ResonanceIDs)

a PeakAssignment stores indices into the lists of spin of proton1 and spin of proton2

example: Spin proton1 – assigns to ( 15, 100 / H 3; H 23 ) Spin label 1 – assign to ( 16, 101 / N 3; N 23 ) Spin proton 2 – assigns to ( 18, 23, 50, 123, 211 / HA 3; QB 10; QD 15; HA 20; H 44 )

PeakAssignments: 1 – 1 ( H 3 – HA 3 ) 1 – 2 ( H 3 – QB 10 ) 1 – 3 1 – 4 1 – 5 2 – 1 ( H 23 – HA 3 ) 2 – 2 ( H 23 – QB 10 ) 2 – 3 2 – 4 2 – 5

the individual assignments are created by "assign_spin" (which is polymorph and calls assigned_labelled_spin if necessary ) the PeakAssignments are subsequently created by an all2all combination between the assignments for proton1 and proton2.

generate all possible initial assignments by chemical shift matching this uses the information about the experiment type: 3D 15N edited, 4D 15N-HN 15N-HN, etc.