Camera: Constraint-based Assignment of Methyl Resonances

Software dependencies

• Python 3.4+ (additional modules below)

  • NumPy
  • networkx
  • Pandas
  • halo
  • tqdm
  • biopython

• cryptominisat5 (bin/env must be able to find this binary)

Preparing inputs

This program is used for assigning methyl resonances measured via NMR. The files required to use this suite of programs are:

1. A crystal structure in PDB format (plus zero or more relaxations)
2. A 2D HMQC peak list in Sparky format
3. A 3D CCH peak list in Sparky format

The first step is to take these files and convert them into a format that is consumable by resonance assignment code.

1. PDB files: the script camera/predict_noes will take as input one or more PDB files and output a .json file. The file that the script outputs will contain a list of all methyls encountered in the PDB files the pairwise distances between all methyls in every structure.

2. HMQC peaklist: the script camera/scripts/hmqc2csv.py takes as input a Sparky HMQC peaklist and outputs a .csv file containing the same information. This suite uses .csv files to store NMR data for ease of parsing and easier human interactivity.

3. 3D CCH peaklist: the script camera/scripts/sparky2csv.py takes as input a Sparky 3D CCH peaklist and outputs a .csv file containing the same information. Make sure that the sparky list you export as input to the script includes 5 columns: a label, the three chemical shifts, and the intensity.

After performing these steps, you should have a .json file containing all information about the known structure, a .csv file describing all 2D peaks, and a .csv file describing all 3D peaks.

In some cases, you will have a 300ms and 50ms CCH experiment. In this case, you want to mark the NOEs which occur in the 50ms experiment as "short" in the 3D .csv file. To do this, run camera/scripts/sparky2csv.py on the 300ms and 50ms peak lists separately to create two .csv files, called 300ms.csv and 50ms.csv. Executing the command camera/mark_short_noes 300ms.csv 50ms.csv -o merged.csv will merge the two files and indicate which NOEs were both observed and symmetrized in the 50ms experiment using a new column entitled short.

Running symmetrization

The first step in assigning the methyl resonances is to process the NOEs. The processing amounts to determining which NOEs are symmetric with each other in order to generate constraints. The script will take as input the structure .json file emitted by camera/predict_noes, the 2D .csv, and the 3D .csv file. The output will be a new 3D .csv file with a column labeled reciprocal which gives all possible reciprocals for each NOE. Issuing the command camera/process_noes hmqc.csv cch.csv structure.json -o processed.csv will emit a list of processed NOEs with reciprocals labeled to processed.csv.

Assigning the HMQC peaks

To run the assignment protocol, use the script assign_signatures. This will take as input the structure .json file emitted by camera/predict_noes, the 2D .csv file, and the processed 3D .csv emitted by camera/process_noes. The script will emit a 2D list with all possible assignments added for each 2D peak. Issuing the command camera/assign_signatures hmqc.csv processed.csv structure.json -o assigned_hmqc.csv will generate a new 2D .csv file called assigned_hmqc.csv with an assignments column with the assignments for each peak.