At the Foresight Institute's [Artificial Intelligence for Scientific Progress: Bringing Digital Control to Physical Matter] (http://ai2016.foresight.org/) workshop in Palo Alto CA in September 2016, a team of participants proposed a project to facilitate the sharing of nanotechnology-related data among nanotechnology researchers, machine learning experts and other interested parties.
The proposal won a prize from the Foresight Institute (as judged by Deepak Srivastava, Bill Goddard, and Ben Goertzel), and took 2nd place, closely behind the winning team, in the "People's Choice" award. It was all relatively informal, but this suggests some widespread interest in the ideas among practitioners.
Here is the project proposal submission for the record.
##Project name Data standards for collaborative nano science and AI
##Team members
- Aaron Virshup
- Neal McBurnett
- Bruce Smith
- Jim Lewis
- Lauren Barghout
- Ruiting Lian
##Project description (50-500 words)
To collaborate on nano science, humans and artificial intelligences need a common language. We will develop a suite of data formats that can relate molecular representations to experimental data, physical properties, synthesis pathways, human cell assay data, and new data types yet to be dreamt of.
We are already working with practitioners at UCSD, MSKCC, Merck, and Autodesk, and using lessons learned from legacy formats such as PDB, mmCIF and CML.
This project builds the digital infrastructure that allows experts pursuing nano-tech, via mutiple paths, to bring digital control to physical matter.
##References
Current related efforts:
- Mosaic: https://github.com/mosaic-data-model
- H5MD: http://nongnu.org/h5md/index.html
- AiiDA: http://www.aiida.net/docs/
An incomplete list of popular formats:
##Data
This format should encode data from:
- Structural databases (PDB, Materials Genome)
- Mechanics codes (ReaxFF, OpenMM, LAMMPS)
- Quantum codes (Gaussian, Quantum Espresso)
- Informatics codes (OpenEye, RDKit)
##Resources Existing informal computational drug discovery collaboration: