Newsletter

Archived Newsletters

UK-QSAR Spring 2025

The UK-QSAR and Cheminformatics Group

Welcome to the Spring 2025 UK-QSAR Newsletter!

 

The Spring UKQSAR & Cheminformatics Society Meeting will be held at the Francis Crick Institute, London, on Tuesday 15th April 2025. The meeting is organised and hosted jointly by MSD and OpenEye, and is free to attend. Registration has now closed but a waiting list is in operation so please continue to register your interest in attending via the registration link here, and if you are already registered but can no longer attend, please let us know so we can free up your slot. . The theme for this meeting is a smorgasbord of science.

Please note that paper copies of the agenda and abstracts will not be provided onsite, so we recommend that you bookmark this page on your device or screenshot the agenda section.

Our newsletter contains more details of the meeting  below including the draft Agenda,  travel and accommodation information, info for  poster presenters and  talk abstracts/pre-reading material, as well as the usual sections on  job openings and upcoming  meetings which may be of interest.  We also welcome a number of  new members to the UKQSAR committee.

Our Autumn Meeting 2025  will be held at AZ in Cambridge, and is provisionally scheduled for 12th November. More details to follow.

Spring Meeting Information

REGISTRATION

Registration is now closed, but you can be added to the waiting list by registering your interest here.

AGENDA

The provisional agenda is:

9:00Open registration, coffee/tea 
9:45Welcome and opening remarksSteve St-Gallay
Session 1 Chair: Val Gillet
10:00A Scientific SmorgasbordPaul Hawkins (OpenEye)
10:30FraGrow Frow Fragments to LLMsPeter EGF Ibrahim (University of Dundee)
11:00Empowering scientists: delivering AI tools through an ELN framework at the enterprise levelTing Qin and Aparna Chandrasekaran (Sygnature Discovery)
11:30Macrocyclic ligand optimisation: Integrating computational approaches with biophysical dataHimani Tandon (Optibrium)
12:00Lunch and poster session 
Session 2 Chair: Mike Bodkin
13:30Harnessing rapid transformer workflows for generative drug designKeishi Kohara (AstraZeneca)
14:00AI/ML Deployment and Orchestration in EvotecWill Pitt (Evotec)
14:30Exploration of synthesis space by application of evolutionary strategiesEmma Armstrong (University of Sheffield)
15:00Break 
Session 3 Chair: Franca Klingler
15:30Benchmarking 3D Structure-Based Molecule GeneratorsNatasha Sanjrani (GSK)
16:00Open Free Energy – Open Source AlchemyJosh Horton (Open Free Energy)
16:30Concluding remarks, poster prize winner announcement and closeDavid Clark

Smoking and vaping are not permitted on site.

PRE-MEETING MEAL

There is a plan for folks local to London or who are arriving the night before the meeting to meet in the Betjeman Arms from 6:00pm The Betjeman Arms | The best pub in St. Pancras Kings Cross. It would be helpful if you could let us know here in advance if you are planning to attend, so we can have a rough idea of numbers.

ABOUT THE CRICK – Steve St-Gallay, MSD

The Francis Crick Institute is an independent charity, established to be a UK flagship for discovery research in biomedicine. The Crick’s mission is discovery without boundaries, so no limits are applied to the direction the research takes, rather the goal is to understand more about how living things work to help improve treatment, diagnosis and prevention of human disease, and generate economic opportunities for the UK.

In the institute more than 2,000 staff and students use their wide-ranging knowledge and expertise to work across disciplines and explore biology at all levels, from molecules through cells to entire organisms.
The Crick is the outcome of a merger between the MRC’s National Institute for Medical Research (NIMR) and CRUK’s London Research Institute (LRI). The new institute was named after the UK scientist Francis Crick in recognition of his contributions to understanding the genetic code, the key to understanding how living things work.  It was established by six founding partner organisations: the Medical Research Council (MRC), Cancer Research UK (CRUK), Wellcome, UCL (University College London), Imperial College London and King’s College London. Scientists began operating in this new, purpose-built institute in early 2017 and it now houses more than 100 research groups. Research groups from NIMR and LRI have been joined by groups seconded from each of our three partner universities, and we have recruited a significant number of newly recruited group leaders.  The institute is governed by a Board of Trustees, comprised of independent members and representatives from each of the founding partners. Groups are funded through core funding from the MRC, CRUK and Wellcome, as well as external research grants.

LOCATION AND TRAVEL 

The best route to the Francis Crick Institute is by rail or London Underground. It is a 2 minute walk from London St Pancras International Station, and 3 minutes from London Kings Cross Station, which can be reached by London Underground (Keeping London moving – Transport for London (tfl.gov.uk)).  London is served by several airports and the Eurostar terminates at St Pancras.

HOTELS

Suggested hotels include:

FOR POSTER PRESENTERS

Abstract submission is now closed.

Poster boards measure 1600 mm x 1600 mm and can display one A0 poster per side. If you are able to specify poster sizes then 2 A1 portrait posters will fit per side, so you can display four per board. Please follow these links to view the posters list and abstracts.

NEW UK-QSAR COMMITTEE MEMBERS

We have had several committee members stepping down from their roles recently due to retirement, new jobs or other reasons (including Nicolas Bosc, Steve Maginn, David Marcus, John Delaney, Nora Aptula and Andrew Leach), so we would like to thank all of them for their effort and commitment to the UKQSAR cause over their many years of service.

We would also like to thank and welcome six new members to the Committee:

Noel O’Boyle (EBI)

Rachael Skyner (OMass Therapeutics)

Daniella Hares (ICR)

Noj Malcolm (Schrodinger)

Lucia Staranova (OpenEye/Cadence)

Baptiste Canault (GSK)

We look forward to working with you all and continuing to provide biannual quality meetings for the computational bioscience community.



Abstracts & Pre-Reading Material

Some speakers have provided the following abstracts and references which might be of interest ahead of their talks. Further abstracts will be added to the website version of this newsletter as they become available.

Structure-based generative molecular design: from 2D or 3D?
Morgan Thomas, Universitat Pompeu Fabra

A Scientific Smorgasbord: Shape and colour: A unifying principle for computational drug discovery
Paul Hawkins, (OpenEye)

The search for an even somewhat widely applicable description of molecular structure and properties and molecular interactions has been ongoing for decades, and many possible solutions have been proposed. Treating molecules as atoms and bonds and assessing their interactions with forcefields has been popular for decades, as it is both a relatively simple representation and has low computational demands. Modelling molecules as nuclei and electrons and using electronic structure methods to calculate interactions is more physically realistic and has become more popular with the rising availability of computational resources. This talk will discuss representing molecules in an extremely simple way – as a shape with embedded chemical features, or colour – to calculate both properties and interactions. This approach carries the appealing conceptual simplicity of forcefields, while being orders of magnitude faster for some applications. We will illustrate the power of the shape and color representation with successful applications across a wide range of tasks in computational drug discovery: target validation, binding site plasticity, virtual screening (both ligand-based and structure-based), and affinity prediction with machine learning


FraGrow Frow Fragments to LLMs
Peter EGF Ibrahim, University of Dundee

Empowering scientists: delivering AI tools through an ELN framework at the enterprise level
Ting Qin and Aparna Chandrasekaran, Sygnature Discovery

Artificial intelligence (AI) has the potential to revolutionize drug discovery, yet its widespread adoption in scientific enterprises faces significant challenges. Key hurdles include ensuring user-friendliness, managing complex workflows, and integrating diverse datasets. To overcome these barriers, we propose a novel framework that incorporates AI into the familiar Electronic Lab Notebook (ELN) paradigm. By embedding AI workflows as ELN protocols and AI job runs as ELN experiments, our approach offers a user-centric, scalable solution that aligns with established scientific practices. This ELN-based framework adheres to FAIR principles – enhancing data findability, accessibility, interoperability, and reusability. By leveraging the intuitive ELN framework, our solution empowers bench scientists to seamlessly access and utilize advanced AI tools, accelerating drug discovery and maximizing the impact of AI investments.

Macrocyclic ligand optimisation: Integrating computational approaches with biophysical data
Himani Tandon, Optibrium

Using the integrated set of computational methods within the BioPharmics Platform, macrocycles can be effectively modelled for lead optimisation. Smaller synthetic macrocyclic ligands or natural products are fully tractable, with modelling of larger peptidic macrocycles benefitting from biophysical NMR or X-ray crystallographic data.

Here, we present two case studies, one involving an active learning approach to lead optimisation of a macrocyclic natural product fungicide, and the other involving optimisation of a macrocyclic inhibitor of the PD-1/PD-L1 interface that was discovered through mRNA library screening. In the first case, neither a protein target structure nor any biophysical data about the lead compound was required to accelerate the optimization process using the QuanSA affinity prediction method. In the second case, with NMR data on the lead compound and a single crystal structure the target protein, acceleration of lead optimisation was demonstrated using a combination of deep conformational search, molecular docking, and careful estimation of bound ligand strain.

Complex peptide macrocycle optimisation: combining NMR restraints with conformational analysis to guide structure-based and ligand-based design – https://optibrium.com/knowledge-base/complex-peptide-macrocycle-optimisation-combining-nmr-restraints-with-conformational-analysis-to-guide-structure-based-and-ligand-based-design/

From UK-2A to florylpicoxamid: active learning to identify a mimic of a macrocyclic natural product https://optibrium.com/knowledge-base/from-uk-2a-to-florylpicoxamid-active-learning-to-identify-a-mimic-of-a-macrocyclic-natural-product/

Harnessing rapid transformer workflows for generative drug design
Keishi Kohara, AstraZeneca

De novo design aims to generate molecules with desirable structural and functional properties in the drug discovery process. REINVENT is a generative model which can build diverse compounds using SMILES notation, using reinforcement learning to guide generation, and is now highly embedded in drug discovery projects at AstraZeneca.

In this talk, I will describe recent updates to the REINVENT4 package, which utilises a new transformer architecture [1]. A deterministic method of sampling the agent latent chemical space with Beam search has been developed, which allows rapid near neighbour searching of a molecule of interest.

I will demonstrate a case study of an example workflow using Beam search in a typical drug design scenario. Multiple agents are used, and the characteristics of each generated output are assessed. The importance of a robust postprocessing pipeline is highlighted, where a Jupyter notebook has been developed and used internally. Postprocessing and triaging methods are used to rationally and rapidly reduce the number of compounds under consideration.

Large-scale synthetic route prediction using AiZynthFinder is used to significantly triage potential designs by synthesisability [2]. This example workflow results in a tractable number of compounds to be assessed by physics-based methods such as free energy perturbation (FEP), leading to suitable synthetic targets.

Saigiridharan, L. et al. AiZynthFinder 4.0: developments based on learnings from 3 years of industrial application. Journal of Cheminformatics 16, 57 (2024).

AI/ML Deployment and Orchestration in Evotec
Will Pitt, Evotec

Loeffler, H. H. et al. Reinvent 4: Modern AI–driven generative molecule design. Journal of Cheminformatics 16, 20 (2024).


In this talk, Will Pitt discusses the deployment and orchestration of AI/ML techniques at Evotec. There will be a brief introduction to how computational chemists operate at the company. Then he will illustrate the use of deep learning latent chemical space in combination with Bayesian optimisation by telling the story of his first project at Evotec. The final part of the presentation will be about how automated workflows, and other improvements have been made since that time. 

Background reading:

Pitt, W., et al, (2025). Real-world applications and experiences of ai/ml deployment for drug discovery, J. Med. Chem. 68, 2, 851–859

Colliandre, L., Muller, C. (2024). Bayesian Optimization in Drug Discovery. In: Heifetz, A. (eds) High Performance Computing for Drug Discovery and Biomedicine. Methods in Molecular Biology, vol 2716.


Exploration of synthesis space by application of evolutionary strategies
Emma Armstrong, University of Sheffield

Benchmarking 3D Structure-Based Molecule Generators
Natasha Sanjrani, GSK

To understand the benefits and drawbacks of 3D combinatorial and deep learning generators, a benchmark was created focusing on the re-creation of important protein-ligand interactions and 3D ligand conformations. Using the BindingMOAD dataset with a hold-out blind set, the deep learning models Pocket2Mol and DiffSBDD were evaluated along with the combinatorial methods AutoGrow4 and LigBuilderV3. It was discovered that deep learning methods fail to generate structurally-valid molecules and 3D conformations whereas combinatorial methods are slow and sample molecules in spaces outside of the required protein active site. The results from this evaluation guide us towards improving deep learning structure-based generators by placing higher importance on structural validity, 3D ligand conformations, and re-creation of important known active site interactions.

Open Free Energy – Open Source Alchemy
Josh Horton, Open Free Energy

Alchemical free energy methods have found tremendous success over the last few decades, becoming key components of drug discovery pipelines. Despite many ongoing innovations in this area, it remains challenging to consistently run free energy campaigns using state-of-the-art tools and best practices. In many cases, doing so requires expert knowledge, and/or the use of expensive closed source software. The Open Free Energy (OpenFE) project (https://openfree.energy/) was created with the aim of addressing these issues. In a joint effort between several academic and industry partners, the project aims to create and maintain reproducible and extensible open-source tools for running large-scale free energy calculations.

In this talk we will summarize the current status of the project and its ecosystem of modular permissively licensed open-source Python tools to routinely set up, calculate, and analyze relative binding free energies (RBFE). We will outline how these various components are brought together in the OpenFE toolkit (https://github.com/OpenFreeEnergy/openfe/) to create a robust framework for carrying out complete free energy campaigns. We conclude by presenting the results of a large-scale pre-competitive collaborative assessment of the default OpenFE RBFE protocol generated by 15 pharmaceutical companies. We evaluate the use of the protocol on both public and blinded private in-house datasets. Our results offer insights into the reliability of alchemical tools, focusing on those offered by OpenFE, and their suitability for use in active drug discovery projects.



Jobs



Upcoming Meetings

Nessa Carson of AZ has been compiling a comprehensive list of scientific conferences, which can be found here.

Notable conferences of particular interest to our readers include:

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.