DEL-GPT: learning the language of DNA-encoded libraries to design focused screening collections
4 affiliations
- 1Department of Chemistry, College of Arts and Sciences
- 2University of Missouri
- 3College of Pharmacy
- 4University of Utah
Abstract
Focused DNA-encoded libraries (DELs) yield higher hit rates and cleaner selection data than billioncompound collections, and are affordable enough to build for a single target. Designing one means choosing a few hundred synthons from a much larger pool, and in the split-and-pool format that choice is collective and irreversible: the library is the full combinatorial product, so each synthon's value depends on every other one chosen alongside it. Brute force is out of reach – there are on the order of 10189 ways to draw 100 synthons from 3,000. We therefore recast the problem as sequence modeling. Treating a synthon as a token and an efficiency-ordered synthon set as a sentence, a generative pretrained transformer trained by next-token prediction learns context-dependent synthon value and composes new sets one synthon at a time. DEL-GPT was trained on one million ranked synthon sets drawn from NADEL, a validated 58,302-member library, with one model per target for five NAD+-dependent enzymes (PARP1, PARP2, PARP10, PARP12, PARP15). Across all five, DEL-GPT libraries outperformed those built from randomly drawn synthons – a demanding baseline, since every NADEL synthon is an expert-selected NAD+ mimetic – and contained two to three times more compounds that experimental affinity selection retained. Quality falls off once generated sets exceed the size range seen in training, but a model trained on longer sequences generated correspondingly larger high-quality sets – a validated recipe for scaling. The method is chemistryagnostic and applies to any combinatorial library over a shared building-block set.
Related
- Exploring the Potential of on-DNA Phthalimide Deprotection for DNA-encoded Libraries
- DNA Directed Chemistry: Scalable Programmed DNA-Encoded Library Synthesis for More Chemically Diverse and Drug-Like Libraries
- On-resin Heck coupling enables small-molecule ligand discovery with covalent peptide-encoded libraries
- AI-enhanced adaptive virtual screening of large libraries for ligand discovery
- Auditing frozen chemical language model embeddings for drug-delivery screening: fingerprint baselines, random-initialization controls, and scaffold-memorization control tasks
BibTeX
@misc{mettu2026delgpt,
title = {{DEL-GPT: learning the language of DNA-encoded libraries to design focused screening collections}},
author = {Akhila Mettu and Naveed Naemi and Raphael Franzini and Dmitri Kireev},
howpublished = {ChemRxiv},
year = {2026},
month = sep,
doi = {10.26434/chemrxiv.15008193/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008193/v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - DEL-GPT: learning the language of DNA-encoded libraries to design focused screening collections
AU - Akhila Mettu
AU - Naveed Naemi
AU - Raphael Franzini
AU - Dmitri Kireev
PY - 2026
DA - 2026/09/02
PB - ChemRxiv
DO - 10.26434/chemrxiv.15008193/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008193/v1
AB - Focused DNA-encoded libraries (DELs) yield higher hit rates and cleaner selection data than billioncompound collections, and are affordable enough to build for a single target. Designing one means choosing a few hundred synthons from a much larger pool, and in the split-and-pool format that choice is collective and irreversible: the library is the full combinatorial product, so each synthon's value depends on every other one chosen alongside it. Brute force is out of reach – there are on the order of 10189 ways to draw 100 synthons from 3,000. We therefore recast the problem as sequence modeling. Treating a synthon as a token and an efficiency-ordered synthon set as a sentence, a generative pretrained transformer trained by next-token prediction learns context-dependent synthon value and composes new sets one synthon at a time. DEL-GPT was trained on one million ranked synthon sets drawn from NADEL, a validated 58,302-member library, with one model per target for five NAD+-dependent enzymes (PARP1, PARP2, PARP10, PARP12, PARP15). Across all five, DEL-GPT libraries outperformed those built from randomly drawn synthons – a demanding baseline, since every NADEL synthon is an expert-selected NAD+ mimetic – and contained two to three times more compounds that experimental affinity selection retained. Quality falls off once generated sets exceed the size range seen in training, but a model trained on longer sequences generated correspondingly larger high-quality sets – a validated recipe for scaling. The method is chemistryagnostic and applies to any combinatorial library over a shared building-block set.
ER -