Widespread false negatives in DNA-encoded library data : how linker effects impair machine learning-based lead prediction
Montoya, Alba L. ; Hogendorf, Adam S. ; Tingey, Steven ; Kuberan, Aadarsh ; Yuen, Lik Hang ; Schüler, Herwig LU
and Franzini, Raphael M.
(2025)
In Chemical Science
16(24).
p.10918-10927
- Abstract
DNA-encoded chemical libraries (DECLs) have become integral to early-stage drug discovery, yielding active compounds and extensive labeled datasets for machine learning (ML)-based prediction of bioactive molecules. However, the information content of DECL selection data remains scarcely explored. This study systematically investigates for the first time the prevalence of false negatives and the influence of the linker in DECL data. Using a focused DECL targeting the poly-(ADP-ribose) polymerases PARP1/2 and TNKS1/2 as a model system, we found that our DECL selections frequently miss active compounds, with numerous false negatives for each identified hit. The presence of the DNA-conjugation linker emerged as a factor contributing to the... (More)
DNA-encoded chemical libraries (DECLs) have become integral to early-stage drug discovery, yielding active compounds and extensive labeled datasets for machine learning (ML)-based prediction of bioactive molecules. However, the information content of DECL selection data remains scarcely explored. This study systematically investigates for the first time the prevalence of false negatives and the influence of the linker in DECL data. Using a focused DECL targeting the poly-(ADP-ribose) polymerases PARP1/2 and TNKS1/2 as a model system, we found that our DECL selections frequently miss active compounds, with numerous false negatives for each identified hit. The presence of the DNA-conjugation linker emerged as a factor contributing to the underdetection of active molecules. This bias toward false negatives compromises the predictive power of DECL data for prioritizing hits, anticipating target selectivity, and training ML models, as determined by analyzing the effects of undersampling and oversampling techniques in learning the PARP2 data. Conversely, the linker's presence in DECLs offers advantages, such as enabling the identification of target-selective protein engagers, even when the underlying molecules themselves may not be selective. These findings highlight the challenges and opportunities of DECL data, emphasizing the need for best practices in data handling and ML model development in drug discovery.
(Less)
- author
- Montoya, Alba L.
; Hogendorf, Adam S.
; Tingey, Steven
; Kuberan, Aadarsh
; Yuen, Lik Hang
; Schüler, Herwig
LU
and Franzini, Raphael M.
- organization
- publishing date
- 2025
- type
- Contribution to journal
- publication status
- published
- subject
- in
- Chemical Science
- volume
- 16
- issue
- 24
- pages
- 10 pages
- publisher
- Royal Society of Chemistry
- external identifiers
-
- scopus:105005756756
- pmid:40395382
- ISSN
- 2041-6520
- DOI
- 10.1039/d5sc00844a
- language
- English
- LU publication?
- yes
- id
- 8075beb0-7ac1-430d-b447-efaab59b1940
- date added to LUP
- 2025-09-24 16:01:42
- date last changed
- 2025-09-25 08:29:39
@article{8075beb0-7ac1-430d-b447-efaab59b1940,
abstract = {{<p>DNA-encoded chemical libraries (DECLs) have become integral to early-stage drug discovery, yielding active compounds and extensive labeled datasets for machine learning (ML)-based prediction of bioactive molecules. However, the information content of DECL selection data remains scarcely explored. This study systematically investigates for the first time the prevalence of false negatives and the influence of the linker in DECL data. Using a focused DECL targeting the poly-(ADP-ribose) polymerases PARP1/2 and TNKS1/2 as a model system, we found that our DECL selections frequently miss active compounds, with numerous false negatives for each identified hit. The presence of the DNA-conjugation linker emerged as a factor contributing to the underdetection of active molecules. This bias toward false negatives compromises the predictive power of DECL data for prioritizing hits, anticipating target selectivity, and training ML models, as determined by analyzing the effects of undersampling and oversampling techniques in learning the PARP2 data. Conversely, the linker's presence in DECLs offers advantages, such as enabling the identification of target-selective protein engagers, even when the underlying molecules themselves may not be selective. These findings highlight the challenges and opportunities of DECL data, emphasizing the need for best practices in data handling and ML model development in drug discovery.</p>}},
author = {{Montoya, Alba L. and Hogendorf, Adam S. and Tingey, Steven and Kuberan, Aadarsh and Yuen, Lik Hang and Schüler, Herwig and Franzini, Raphael M.}},
issn = {{2041-6520}},
language = {{eng}},
number = {{24}},
pages = {{10918--10927}},
publisher = {{Royal Society of Chemistry}},
series = {{Chemical Science}},
title = {{Widespread false negatives in DNA-encoded library data : how linker effects impair machine learning-based lead prediction}},
url = {{http://dx.doi.org/10.1039/d5sc00844a}},
doi = {{10.1039/d5sc00844a}},
volume = {{16}},
year = {{2025}},
}