D2ML: Models for Direct-to-Biology Small Molecule Screening
Abstract
Direct-to-biology (D2B) screening enables rapid testing of newly synthesized molecules without purification. However, uncertain concentrations in crude mixtures and limited purified-compound measurements complicate the use of these data in machine learning. Here we show that models tailored to the structure of D2B data can improve prediction and guide the discovery of more potent and selective compounds. Our assay-aware models predict activity from single-point screens, relative potency across targets from crude dose-response assays, and absolute potency from purified-compound measurements. For crude assays, potency ratios between targets tested from the same well can retain selectivity information despite uncertain concentrations. A joint model that shares information across stages and balances their contributions to training improved predictions over separate neural networks. The largest gain was in crude dose-response selectivity prediction, where rank correlation increased from 0.27 to 0.40. In a separate experiment, independently trained models guided the selection of 84 new compounds to improve ROS1 potency while limiting activity against the anti-target TRKB. Synthesis and testing identified compounds that extended the measured Pareto front, achieving better combinations of potency and selectivity than those in the training data. The most selective new compound achieved a ROS1 IC50 of 14.7 nM and 233-fold selectivity over TRKB. By comparison, the most selective training compound had a ROS1 IC50 of 20.3 nM and 79-fold selectivity. ATLAS provides the experimental setting for this approach, supporting synthesis and screening of 1,000-10,000 model-selected compounds in two- to three-week cycles. Together, these results show how assay-aware learning can use the full D2B assay sequence to guide multi-objective molecular optimization.
Related
- Quantifying structure and performance diversity for sets of small molecules comprising small-molecule screening collections
- A high-throughput screening dataset of small-molecule inhibitors across human DNA glycosylases
- Computational assay of hERG ion permeation and its modulation by small-molecule compounds
- A small molecule inhibitor of human ARV1 sterol-binding protein blocks lipid binding
- Deep reinforcement learning-driven discovery of a MsbA-targeted small-molecule antibiotic for the treatment of Acinetobacter baumannii infection
BibTeX
@misc{nigam2026d2ml,
title = {{D2ML: Models for Direct-to-Biology Small Molecule Screening}},
author = {AkshatKumar Nigam and Neri Amara and Hillary Dequina and Dyana Kenanova and Seth Malmersjo and Fiorella Ruggiu and Raymond V. Fucini and Jack Sadowsky and Stig K. Hansen and Imran S. Haque},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.41070},
eprint = {2610.41070},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.41070},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - D2ML: Models for Direct-to-Biology Small Molecule Screening
AU - AkshatKumar Nigam
AU - Neri Amara
AU - Hillary Dequina
AU - Dyana Kenanova
AU - Seth Malmersjo
AU - Fiorella Ruggiu
AU - Raymond V. Fucini
AU - Jack Sadowsky
AU - Stig K. Hansen
AU - Imran S. Haque
PY - 2026
DA - 2026/10/29
PB - arXiv
DO - 10.48550/arxiv.2610.41070
UR - https://arxiv.org/abs/2610.41070
AB - Direct-to-biology (D2B) screening enables rapid testing of newly synthesized molecules without purification. However, uncertain concentrations in crude mixtures and limited purified-compound measurements complicate the use of these data in machine learning. Here we show that models tailored to the structure of D2B data can improve prediction and guide the discovery of more potent and selective compounds. Our assay-aware models predict activity from single-point screens, relative potency across targets from crude dose-response assays, and absolute potency from purified-compound measurements. For crude assays, potency ratios between targets tested from the same well can retain selectivity information despite uncertain concentrations. A joint model that shares information across stages and balances their contributions to training improved predictions over separate neural networks. The largest gain was in crude dose-response selectivity prediction, where rank correlation increased from 0.27 to 0.40. In a separate experiment, independently trained models guided the selection of 84 new compounds to improve ROS1 potency while limiting activity against the anti-target TRKB. Synthesis and testing identified compounds that extended the measured Pareto front, achieving better combinations of potency and selectivity than those in the training data. The most selective new compound achieved a ROS1 IC50 of 14.7 nM and 233-fold selectivity over TRKB. By comparison, the most selective training compound had a ROS1 IC50 of 20.3 nM and 79-fold selectivity. ATLAS provides the experimental setting for this approach, supporting synthesis and screening of 1,000-10,000 model-selected compounds in two- to three-week cycles. Together, these results show how assay-aware learning can use the full D2B assay sequence to guide multi-objective molecular optimization.
ER -