Hierarchical Resolution Calculations of Biomolecular Information

Overview This repository holds source code and results used in Kinz-Thompson and Ray, DOI: 10.1101/2023.10.17.562748. Code is written in Python, and running it requires the following libraries: `numpy`, `scipy`, `matplotlib`, `h5py`, `numba`, and `requests`. Data is saved in plain text, or self-described HDF5 format. Calculations 1. Sensitivity Simulations This calculation simulates images of a proto-molecule at different levels of peak signal-to-noise ratios (PSNR), and then analyzes the images using hierarchical resolution calculations. The results are shown in Fig. 3. Source Files: Filename Format Description resolution_criterion_fxns.py Python A library of functions required for the hierarchical resolution calculation. The calculation uses this file. It is not something that is run by itself. plot_snr_curves.py Python Simulates images, analyzes them, saves the results in `sensitivity_simulations.hdf5`, and then makes a plot of <P> vs. normalized PSF width as a function of PSNR. plot_snr_examples.py Python Makes plots of the cross-sectional profile and example simulated images. Data Files: Filename Format Description sensitivity_simulations.hdf5 HDF5 Holds all data used to make the plot of <P> vs. normalized PSF width as a function of PSNR. This data is plotted in Fig. 3. 2. Nearest Neighbor Distances This calculation searches through the majority of the structures deposited in the Protein Data Bank (PDB) to investigate the distribution of distances from residue to nearest residue. This results are shown in Fig. S1. Note: downloading this many structures from the PDB requires a significant amount of storage space. Source Files: Filename Format Description mmcif.py Python A library without much overhead to load in compessed .cif.gz files into a custom `molecule` class where residues can be located calculate_distances.py Python Queries the PDB for relevant structures, downloads all the .cif.gz files in the PDB using rsync, calculates the nearest-neighbor distance for each residue in each structure using multiprocessing, collects all the results into a single HDF5 file (`distance_histograms.hdf5`), makes a composite histogram of all results (`distance_histogram.txt`), and finally plots the histogram. sigma0_calc.py Python Radially integrates a 3D Gaussian to find the radius where half of the density is below a cutoff and half is above. Data Files: Filename Format Description distance_histograms.hdf5 HDF5 Each entry key is a PDBID, and the associated dataset is the histogram of nearest-neighbor distances. Also includes an entry 'hx' which is the left-bind edge of the histograms (common to all). distance_histogram.txt ASCII Text A composite histogram of all the histograms in `distance_histograms.hdf5`. Can be loaded using numpy's `np.loadtxt()`. This data is plotted in Fig. S1.