Global reactivity models are impactful in industrial synthesis applications
Abstract Artificial Intelligence is revolutionizing many aspects of the pharmaceutical industry. Deep learning models are now routinely applied to guide drug discovery projects leading to faster and improved findings, but there are still many tasks with enormous unrealized potential. One such task i...
Main Authors: | , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
BMC
2023-02-01
|
Series: | Journal of Cheminformatics |
Online Access: | https://doi.org/10.1186/s13321-023-00685-0 |
_version_ | 1797863713200406528 |
---|---|
author | Paulo Neves Kelly McClure Jonas Verhoeven Natalia Dyubankova Ramil Nugmanov Andrey Gedich Sairam Menon Zhicai Shi Jörg K. Wegner |
author_facet | Paulo Neves Kelly McClure Jonas Verhoeven Natalia Dyubankova Ramil Nugmanov Andrey Gedich Sairam Menon Zhicai Shi Jörg K. Wegner |
author_sort | Paulo Neves |
collection | DOAJ |
description | Abstract Artificial Intelligence is revolutionizing many aspects of the pharmaceutical industry. Deep learning models are now routinely applied to guide drug discovery projects leading to faster and improved findings, but there are still many tasks with enormous unrealized potential. One such task is the reaction yield prediction. Every year more than one fifth of all synthesis attempts result in product yields which are either zero or too low. This equates to chemical and human resources being spent on activities which ultimately do not progress the programs, leading to a triple loss when accounting for the cost of opportunity in time wasted. In this work we pre-train a BERT model on more than 16 million reactions from 4 different data sources, and fine tune it to achieve an uncertainty calibrated global yield prediction model. This model is an improvement upon state of the art not just from the increase in pre-train data but also by introducing a new embedding layer which solves a few limitations of SMILES and enables integration of additional information such as equivalents and molecule role into the reaction encoding, the model is called BERT Enriched Embedding (BEE). The model is benchmarked on an open-source dataset against a state-of-the-art synthesis focused BERT showing a near 20-point improvement in r2 score. The model is fine-tuned and tested on an internal company data benchmark, and a prospective study shows that the application of the model can reduce the total number of negative reactions (yield under 5%) ran in Janssen by at least 34%. Lastly, we corroborate the previous results through experimental validation, by directly deploying the model in an on-going drug discovery project and showing that it can also be used successfully as a reagent recommender due to its fast inference speed and reliable confidence estimation, a critical feature for industry application. |
first_indexed | 2024-04-09T22:40:12Z |
format | Article |
id | doaj.art-21f6f1caf46a444b96b2c04825d4b4b1 |
institution | Directory Open Access Journal |
issn | 1758-2946 |
language | English |
last_indexed | 2024-04-09T22:40:12Z |
publishDate | 2023-02-01 |
publisher | BMC |
record_format | Article |
series | Journal of Cheminformatics |
spelling | doaj.art-21f6f1caf46a444b96b2c04825d4b4b12023-03-22T12:13:30ZengBMCJournal of Cheminformatics1758-29462023-02-0115111110.1186/s13321-023-00685-0Global reactivity models are impactful in industrial synthesis applicationsPaulo Neves0Kelly McClure1Jonas Verhoeven2Natalia Dyubankova3Ramil Nugmanov4Andrey Gedich5Sairam Menon6Zhicai Shi7Jörg K. Wegner8In-Silico Discovery and External Innovation (ISDEI), Janssen Research & Development, Janssen Pharmaceutica N.VDiscovery Chemistry LJ, Janssen Research & Development, Janssen Pharmaceutica N.VIn-Silico Discovery and External Innovation (ISDEI), Janssen Research & Development, Janssen Pharmaceutica N.VIn-Silico Discovery and External Innovation (ISDEI), Janssen Research & Development, Janssen Pharmaceutica N.VIn-Silico Discovery and External Innovation (ISDEI), Janssen Research & Development, Janssen Pharmaceutica N.VSoftware CountryPharma R&D Information Tech, Janssen Research & Development, Janssen Pharmaceutica N.VDiscovery Chemistry LJ, Janssen Research & Development, Janssen Pharmaceutica N.VIn-Silico Discovery and External Innovation (ISDEI), Janssen Research & Development, Janssen Pharmaceutica N.VAbstract Artificial Intelligence is revolutionizing many aspects of the pharmaceutical industry. Deep learning models are now routinely applied to guide drug discovery projects leading to faster and improved findings, but there are still many tasks with enormous unrealized potential. One such task is the reaction yield prediction. Every year more than one fifth of all synthesis attempts result in product yields which are either zero or too low. This equates to chemical and human resources being spent on activities which ultimately do not progress the programs, leading to a triple loss when accounting for the cost of opportunity in time wasted. In this work we pre-train a BERT model on more than 16 million reactions from 4 different data sources, and fine tune it to achieve an uncertainty calibrated global yield prediction model. This model is an improvement upon state of the art not just from the increase in pre-train data but also by introducing a new embedding layer which solves a few limitations of SMILES and enables integration of additional information such as equivalents and molecule role into the reaction encoding, the model is called BERT Enriched Embedding (BEE). The model is benchmarked on an open-source dataset against a state-of-the-art synthesis focused BERT showing a near 20-point improvement in r2 score. The model is fine-tuned and tested on an internal company data benchmark, and a prospective study shows that the application of the model can reduce the total number of negative reactions (yield under 5%) ran in Janssen by at least 34%. Lastly, we corroborate the previous results through experimental validation, by directly deploying the model in an on-going drug discovery project and showing that it can also be used successfully as a reagent recommender due to its fast inference speed and reliable confidence estimation, a critical feature for industry application.https://doi.org/10.1186/s13321-023-00685-0 |
spellingShingle | Paulo Neves Kelly McClure Jonas Verhoeven Natalia Dyubankova Ramil Nugmanov Andrey Gedich Sairam Menon Zhicai Shi Jörg K. Wegner Global reactivity models are impactful in industrial synthesis applications Journal of Cheminformatics |
title | Global reactivity models are impactful in industrial synthesis applications |
title_full | Global reactivity models are impactful in industrial synthesis applications |
title_fullStr | Global reactivity models are impactful in industrial synthesis applications |
title_full_unstemmed | Global reactivity models are impactful in industrial synthesis applications |
title_short | Global reactivity models are impactful in industrial synthesis applications |
title_sort | global reactivity models are impactful in industrial synthesis applications |
url | https://doi.org/10.1186/s13321-023-00685-0 |
work_keys_str_mv | AT pauloneves globalreactivitymodelsareimpactfulinindustrialsynthesisapplications AT kellymcclure globalreactivitymodelsareimpactfulinindustrialsynthesisapplications AT jonasverhoeven globalreactivitymodelsareimpactfulinindustrialsynthesisapplications AT nataliadyubankova globalreactivitymodelsareimpactfulinindustrialsynthesisapplications AT ramilnugmanov globalreactivitymodelsareimpactfulinindustrialsynthesisapplications AT andreygedich globalreactivitymodelsareimpactfulinindustrialsynthesisapplications AT sairammenon globalreactivitymodelsareimpactfulinindustrialsynthesisapplications AT zhicaishi globalreactivitymodelsareimpactfulinindustrialsynthesisapplications AT jorgkwegner globalreactivitymodelsareimpactfulinindustrialsynthesisapplications |