reference_handler

A Python package that creates a consolidated list of references (BibTex, RIS) of scientific material used in a Python package.

Motivation

Authors of modeling tools (scientific software, method or model developers) currently find it difficult to get fair attribution to their software and scientific work. Scientific software developers create tools that do not always get cited, as a typical citable scientific article might not exist. Although new mechanisms for more directly citing software have been recently created (i.e. Zenodo), they have not gained widespread use in the community of computational molecular science.

Method and model developers face a similar problem. Consider for instance a molecular dynamics practitioner who needs to develop a new force field using a specialized quantum mechanical calculation. The practitioner might not be fully familiar with the correct references to use for the quantum mechanical software and methods. To correctly include the references, the model developer might spend a considerable amount of time finding out which papers to cite. This problem is exponentially exacerbated in work that involves complex workflows running hundreds of different software components that might involve different areas of computational science.

Goal

The goal of the reference handler is to provide an easy mechanism for developers to record the appropriate references so that users of tools can provide a complete set of citations for a particular run of the software in a form convenient for the user.

The output of reference handler is a consolidated list of references (BibTeX, RIS) to go into the paper with as little effort as possible on the users part.

Audience

1. Method and scientific software developers.

2. Computational molecular science practitioners

Package overview

The reference handler is comprised of the following:

1. Central SQLite3 database. It contains two tables. The first is named citation and holds the essential data associated to each unique citation, such as the raw citation text and its ID number. The second is the context table. It contains information about the context in which a given citation was used. For instance, the function where the citation was used or the number of times the citation was "mentioned" by any function of your Python package.

Each citation can have many contexts, yielding a one-to-many relationship between the two tables.

**2. Functions to ease the interaction with central database. ** Examples are a function to cite a desired reference or a function to dump the contents of the database into a .bib file for subsequent compliation using BibTeX.

Minimal example

import reference_handler

lj_citation = """
@article{lj1924,
author = {J. E. Jones  and Sydney Chapman },
title = {On the determination of molecular fields. \&\#x2014;II. From the equation of state of a gas},
journal = {Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathematical and Physical Character},
volume = {106},
number = {738},
pages = {463-477},
year = {1924},
doi = {10.1098/rspa.1924.0082},
}
"""

def lennard_jones(sigma, epsilon, rij):
    rf.cite(raw=lj_citation, alias='lj_citation', module='lennard_jones',
        level=1, note='The first version of the Lennard-Jones potential')
    sig_by_r6 = (sigma / rij ** 6)
    sig_by_r12 = (sig_by_r6 ** 2)
    return 4.0 * epsilon * (sig_by_r12 - sig_by_r6)

rf = reference_handler.Reference_Handler('database.db')

sigma = 3.54
epsilon = 98.0
rij = 20.0

lj_energy = lennard_jones(sigma, epsilon, rij)

rf.dump(outfile='bibliography.bib')

Installation

Documentation

Contribute

Citation

Copyright

Copyright (c) 2019, MolSSI

Acknowledgements

Project based on the Computational Molecular Science Python Cookiecutter version 1.0.