ACS Fall, August 25-29, 2019 | ACD/Labs
ACD Labs Logo

ACS Fall

August 25-29, 2019
San Diego Convention Center, San Diego, CA, USA

Oral Presentation Schedule

SUNDAY, AUG. 25TH, 3:40 PM
Session: CINF: Division of Chemical Information
Location: Grand Ballroom D, Omni San Diego Hotel

Chemical nomenclature from books to computers—ACD/Name and IUPAC Division VIII
Andrey Yerin

Abstract: The IUPAC, responsible for nomenclature development and celebrating its centennial this year, has published several volumes of specific rules. But in over 200 years of development, the nomenclature has grown too complex to learn and apply.

Algorithmic name generation entered the market in the early 1990s, with ACD/Labs among the first software providers with ACD/Name. The real challenge for programming was taking the multitude of nomenclature rules and converting them into computer algorithms. Since the early stages of our software’s development, ACD/Labs has been invited to work with the IUPAC Commission on the Nomenclature of Organic Chemistry and later the Chemical Nomenclature and Structure Representation Division. This involvement in nomenclature projects allowed the development team behind Name to learn nomenclature in greater detail, and helped turn those huge printed volumes into reliable algorithms of name generation.

At the same time our work developing nomenclature algorithms allowed us to detect areas that lacked the necessary criteria, and propose procedures that were eventually included in the current IUPAC recommendations. With the implementation of the Preferred IUPAC Name (PIN) concept in Name algorithms, a number of name errors were identified in the 2013 version of the IUPAC Blue Book. Corrections to these errors are expected to be published in the Errata for this book; once again proving that even nomenclature experts make mistakes, and can benefit from algorithmic name generation.

The past 25 years of ACD/Name development showcases the mutual benefits for both chemical nomenclature as a field of study, and software developers alike. While name generation tools from several vendors are now available and heavily used for name generation in various electronic media, algorithmic nomenclature development is far from complete, and requires further collaboration of nomenclature bodies and software vendors to ensure high quality names for all classes of chemical substances.

MONDAY, AUG. 26TH, 4:10 PM
Session: ORGN: Artificial Intelligence in Organic Synthesis
Location: Room 8, San Diego Convention Center
Abstract #: ORGN 277

Chemical reaction data reuse: Preparing ELN data for analytics and prediction
Scott Harrison, GSK

Abstract: In the field of Computer Aided Synthesis Planning (CASP), it has been hypothesized that certain classes of prediction will require high-quality reaction datasets that include reaction successes and failures. Access to this data is particularly important in attempts to predict the probability of a transformation’s technical success. Academic and patent literature contain a rich corpus of transformations that have been successfully reduced to practice, however unproductive transformations are rarely reported. Most pharmaceutical R&D organizations have deployed electronic laboratory notebooks (ELNs) to provide a mechanism to capture successful and unsuccessful chemical reactions. These tools were largely designed for maximum flexibility in capturing intellectual property, with less emphasis on data reuse in reaction analytics and machine learning-powered chemical reaction predictions. This talk describes the challenges we faced in curating our ELN chemical reaction data, steps we are taking to improve chemical reaction data capture and curation, and preliminary attempts to train machine learning models of potential utility to synthetic organic chemists.