loco-hd 0.1.1

Implementation of the LoCoHD metric for quantitative protein structure and substructure comparison

Welcome to LoCoHD!

LoCoHD (Local Composition Hellinger Distance) is a metric for comparing protein structures. It can be used for one single structure-structure comparison, for the comparison of multiple structures inside ensembles, or for the comparison of structures inside an MD simulation trajectory. It is also a general-purpose metric for labelled point clouds with variable point counts. In contrast to RMSD, the TM-score, lDDT, or GDT_TS, it is based on the measurement of local composition differences, rather than of the Euclidean deviations.

Where can I read about it?

The theory behind LoCoHD was published here.

How can I install it?

From PyPI

With pip, it is easy to add LoCoHD to your packages:

pip install loco_hd

Building from source

To build LoCoHD from source, first you need to install Rust to your system. You also need Python3, pip, and the package Maturin. Both Rust and Maturin can be installed with the following one-liners:

curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

pip install maturin

Next, clone the repository and enter it:

git clone https://github.com/fazekaszs/loco_hd && cd loco_hd

Run Maturin to install LoCoHD into your active environment:

maturin develop

And you are done!

How can I use it?

LoCoHD was intended to be used within Python scripts, mostly through BioPython as the main .pdb file reader. It is also possible to use it with other protein/molecular structure readers, but the user has to write the appropriate parser that converts the information within the file into the information required for LoCoHD. An example for this can be found here, where the structures come from a molecular dynamics trajectory and parsing is achieved by MDAnalysis.