MPrint OKN

Accelerating molecular discovery by harnessing the data revolution
We are OKN-instrument builders

About MPrint OKN

Open Knowledge Network

Facebook, Amazon, Apple, Netflix, and Alphabet offer customers powerful tools in exchange for their data. Machine learning algorithms analyze the data and the output provides value to another set of customers. Our Open Knowledge Network follows a similar model where users are incentivized to share data, because they gain access to powerful tools that help them maximize the value of their data. The MPrint-OKN will deliver valuable tools for ingesting, curating, modeling, analyzing and visualizing molecular data.


The MPrint-OKN will span the gap between data science/machine learning and chemical science. MPrint OKN will unite disciplines, promote partnerships, and provide training to those who wish to accelerate their research with data science/machine learning.

What is an MPrint?

The quantum mechanical wavefunction(s) that define molecular performance are not captured by existing representations such as line drawings, input line-entries, and extended connectivity fingerprints.

Three different approaches will be used to obtain the quantum mechanical data be used to form an MPrints: (1) calculations using third party software, (2)  repositories of existing calculations, or (3) the ANI-1 potential.

These data will be used to form MPrints of three different sizes: (1) the full wavefunction as approximated by DFT calculations; (2) an ensemble of molecular descriptors calculated from the wavefunction such as polarizability and orbital information; and (3) further abstractions of quantities such as volume and solvent accessible surface area.

How can an MPrint be used?

Our hypothesis is that sufficiently resolved models of molecular performance enabling design will only emerge from machine learned correlations that use more developed molecular representations (i.e. MPrints) and performance.​

The MPrint-OKN will enable community members to create a set of machined-learned models against network measurement data, eg. technical specification databases, data leveraged from US Government or other public databases, or contributed validated performance data sets stored within the OKN or elsewhere. The resulting model(s) will aid design of new materials by using an auto encoder to bridge continuous space back to discrete molecule space for example. In addition to leveraging databases to identify networks, patterns, and correlations.



D. Tyler McQuade, Ph.D.


Chief Technical Officer

Department of Chemical and Life Science Engineering

Medicines for All Institute

Virginia Commonwealth University


Tyler serves as Program Director.

James K. Ferri, Ph.D.

Professor and Associate Chair

Department of Chemical and Life Science Engineering

Virginia Commonwealth University

James leads the Measurement

and Applications performance data working group.

Carol Parish Ph.D.


Floyd D. and Elisabeth S. Gottwald Chair

Department of Chemistry

University of Richmond

Carol leads the Quantum Mechanics working group.

Adrian Roitberg Ph.D.


Department of Chemistry

University of Florida

Adrian leads the Data Science working group.