The Liverpool Materials Discovery Server: A suite of computational tools for the collaborative discovery of materials

Abstract

The discovery of new materials often requires collaboration between experimental and computational chemists. Web based platforms allow more flexibility in this collaboration by giving access to computational tools without the need for access to computational researchers. We present Liverpool Materials Discovery Server (lmds.liverpool.ac.uk), one such platform which currently hosts six state of the art computational tools in an easy to use format. We describe the development of this platform, highlighting the advantages and disadvantages the methods used. In addition, we provide source code, and setup scripts to enable other research groups to create similar platforms, to promote collaboration both within and between research groups.

The Liverpool materials discovery server: a suite of computational tools for the collaborative discovery of materials

“…Difficulties in applying computational methods in experimental research are discussed, as are the barriers for deployment of such methods to cloud platforms. Examples of the tools discussed in this manuscript have been provided, 69,88 as well as scripts setting these up with an Nginx reverse proxy server, and Python process manager, 78 enabling other researchers to reproduce this tool chain and share their own methods, either internally or on the open internet.…”

“…The modeling of heat capacity was carried forward using the SciPy library. 94 Each of the web applications are implemented using Flask, with the gunicorn process manager used to spawn Flask processes, examples of these implementations have been provided, 69,88 including a basic Flask app, which can be used as templates for other researchers. 95 VMs are run through VMWare vSphere ESXi 7, 96 with Nginx used to serve HTTP responses to users and route URLs to each VM.…”

“…In order to suit a variety of potential use cases, this tool can take a variety of inputs, being; two equal length comma separated lists (one of metals, one of linkers); one linker and a list of metal symbols (or vice versa); or a single metal and linker combination can be input. Due to limitations in the training set of this model, only certain metals are supported, a list of these can be found in the associated code repository, 69 and is linked to on the application itself.…”

“…While ample learning materials for such skills exist online, this adds a barrier to the launch of new web applications. While styling presented here 65,69,88 can be used as a template, it is likely research groups will wish to develop their own aesthetic with which to brand their website. While heavily adjusted, the front end seen here was outsourced commercially.…”