All datasets used to test and characterize camera are stored within subdirectories here. Each data directory contains acqus files generated from Bruker Topspin, and a scripts subdirectory that holds all necessary conversion, reconstruction and processing steps for that dataset.
Within each subdirectory here, a scripts directory exists that holds all relevant reconstruction and processing procedures for that dataset. There are several scripts for each dataset, each of which has a well-defined role:
The fid.sh script converts raw Bruker NUS data in the ser file into the nmrPipe format. For reconstructing raw Varian NUS data using camera, only this script needs modification.
The proc0.sh script processes the direct (uniformly sampled) dimension of the nmrPipe-format time-domain NUS data, but does not prepare the data for reconstruction. This script should be used to determine the direct dimension phase correction prior to reconstruction.
The proc1.sh script performs the exact same direct-dimension processing as proc0.sh, but also prepares the processed NUS data for reconstruction by transposition. The transposition step places the direct dimension last in the dimension ordering, so that the indirect dimensions may be read in-order for reconstruction.
The procR.sh script reconstructs the indirect (nonuniformly sampled) dimensions of the partially Fourier-transformed data.
The proc2.sh (2D), proc23.sh (3D), and proc234.sh (4D) scripts process the reconstructed indirect dimensions using standard techniques.
The conv.sh script converts the fully reconstructed and processed dataset into nmrView format for visualization in other software packages, like CCP-NMR.
To build (i.e. reconstruct and process) all the datasets, you can just run make from this directory. To build an individual dataset, simply run make from its scripts subdirectory. For example, to build the HNCO dataset, run:
cd hnco/scripts
makeRaw serial files are not stored in this GitHub repository, for the sake of minimizing the disk storage required by the camera source tree. However, the scripts/Makefile within each data subdirectory contains a Dropbox link to its respective ser file. Thus, running make here or in any scripts subdirectory will automatically download all required serial files.
The nhsqc-gb1 directory contains a 1H-15N HSQC originally provided by the Wagner group to demonstrate the [hmsIST] (http://gwagner.med.harvard.edu/intranet/hmsIST/) program. It was collected using Poisson-gap sampling at 10% sparsity on a 512-point hypercomplex grid.
The nhsqc-ubq-sigma directory holds a 1H-15N HSQC of ubiquitin, originally collected to compare the performance of Poisson-gap, sine-gap and sine-burst sampling in one dimension:
Worley, B., Powers, R., Deterministic Multidimensional Nonuniform Gap Sampling, Journal of Magnetic Resonance, 2015, 261: 19-26.
The data was collected using a sine-gap schedule at 5% sparsity on a 1k-point hypercomplex grid. This folder contains scripts to reconstruct the data in constant-aim (i.e. constant-sigma) mode.
The nhsqc-ubq-lambda directory has the same dataset of ubiquitin, but with scripts to reconstruct the data in constant-lambda mode. The value of lambda used in these scripts was determined by examining the resulting Lagrange multipliers from the constant-aim dataset in the nhsqc-ubq-sigma directory.
The chsqc-ca directory houses a semi-selective 1H-13C HSQC of ubiquitin, with evolution times set to select for alpha carbons. The data was collected at 5% sparsity on a 2048-point grid using sine-gap sampling.
This directory additionally demonstrates the use of J-coupling deconvolution in camera to produce singlet spectra when coupling patterns are known.
The chsqc-cacb directory holds the same 1H-13C HSQC as the chsqc-ca directory, but with evolution times set to select for both alpha and beta carbons. No deconvolution is performed for this dataset.
The hnco directory contains an HNCO of ubiquitin, acquired for reference purposes for the [UNL RIF] (http://chem.unl.edu/research-instrumentation). It was collected using a sine-burst schedule at 5% sparsity on a 48x128-point hypercomplex grid.
The hnco-ext directory holds the same HNCO dataset, but with slightly different reconstruction parameters. Specifically, this is an testbed for how camera behaves when asked to simultaneously perform MaxEnt reconstruction and extrapolation. Thus, the original 48x128-point dataset is extrapolated into a 128x256-point grid.
The noesy directory holds an HCCH-NOESY (HSQC-NOESY-HSQC) of ubiquitin, collected specifically to implement and test camera on four-dimensional data. It was collected using a subrandom exponentially weighted (with 2x bias) schedule at 0.8% sparsity on a 64x64x64-point hypercomplex grid.