SPICE: A Dataset for Training Machine Learning Potentials

Default Boolean Operator: Terms in a search string will be connected by the AND operator, EXAMPLE: searching genomic melanoma will produce the same results as genomic AND melanoma .
AND, OR, NOT: Combine terms in a search with the boolean operators AND (to add specificity), OR (to broaden), or NOT (to eliminate a term). Operators must be capitalized (all other search terms are not case-sensitive).
Wild Card: Add wild card symbols * (open-ended) and ? (single character) to truncate terms. EXAMPLE: searching gen* shows results for 'gene', 'genetic', 'genomic, etc. Searching gene? returns results for 'gene' or 'genes' but not for 'genome'.
Phrase Searching: Enclose search in quotation marks "keyword or search term" to search for an exact phrase. EXAMPLE: "transcription factors" returns results for records where transcription is immediately followed by the word factors .
Diacritics: Words containing diacritics translate those characters to non-diacritic form. EXAMPLE: massague returns results for Massague&#769 .
Filter by: Located on the homepage (left panel), use filters to search for records or narrow the search. From search results page, filters can be removed by clicking on the term at the top of the page.

UID: 11387

Author(s): Eastman, Peter

Description: Description from GitHub:

"This repository contains scripts and data files used in the creation of the SPICE dataset. It does not contain the dataset itself. It does contain a collection of QM data for training potential functions."

Subject(s)

Access via GitHub
Access Restrictions: Free to All
Access Instructions: Available for viewing and download via Github under an MIT License.
Associated Publications: Eastman P, Behara P, Dotson D, Galvelis R, Herr J, Horton J, Mao Y, Chodera J, Pritchard B, Wang Y, De Fabritiis G, Markland T. SPICE, a dataset of drug-like molecules and peptides for training machine learning potentials. Scientific Data. 2023; 10:11.
Software Used: OpenFF

OpenMM

openmmforcefields

Psi4
Open-Source Quantum Chemistry – an electronic structure package in C++ driven by Python

QCPortal
a front-end to a QCFractal server which allows the querying, visualization, manipulation of hosted data.

RDKit
Dataset Format(s): YAML, gzip, PDB, PY, MD, TDD, HDF5
Dataset Size: 757 MB

Do you have or know of a dataset that should be added to the catalog? Let us know!