1859 appoints Dr. Helena Marsh as Chief Scientific Officer — read the announcement

1859 The Origin of Next Generation Medicines

Publications

Methods, benchmarks and results.

We publish the assays, the model architectures and the negative results. Where a method is used to prioritise synthesis in a programme, it appears here first.

2026

Pico-scale biochemical screening for structure–activity model training

Marsh H, Okonjo A, Reyes D, Lindqvist S, Agrawal S

Nature Chemical Biology, 2026

2026

Uncertainty-driven library design reduces the cost of lead identification

Reyes D, Feldman J, Marsh H, Agrawal S

Journal of Medicinal Chemistry, 2026

2025

Assay acceptance criteria as a source of bias in machine-learned potency models

Okonjo A, Marsh H, Whitfield R

SLAS Discovery, 2025

2025

Transfer of selectivity models across a kinase family with limited wet-lab data

Feldman J, Okonjo A, Lindqvist S

Preprint, 2025

2024

Practical limits of combinatorial library size in plate-based synthesis

Whitfield R, Marsh H, Reyes D

ACS Combinatorial Science, 2024

2023

Industrialising empirical data generation for small molecule discovery

Lindqvist S, Agrawal S, Marsh H

Drug Discovery Today, 2023

A plate heat map of measured activity beside a fitted dose-response curve with measured points

Data & methods

Assay protocols, model architectures and evaluation splits are made available to academic collaborators on request, subject to the terms of the relevant research agreement. Where a dataset can be released without compromising an active programme, we release it.

  • Assay protocols including acceptance windows and control specifications
  • Model cards describing training data, splits and known failure modes
  • Benchmark definitions, fixed in advance of evaluation
  • Negative results, published alongside the positive ones

Request methods documentation