nextAID³

(Kopie 1)

On 27 June 2024, Dr David B. Konrad (LMU) will give a lecture on "Unlocking the Potential of Late-Stage Functionalization for Medicinal Chemistry by Integrating High-Throughput Experimentation with Geometric Deep Learning". The lecture will start at 3 pm in the HIPS (seminar room 0.27, building E8 1, Saarbrücken campus).

Abstract

Structural novelty and complexity render the synthesis of chemical target structures
challenging when aiming to access structure-activity relationships (SARs) in medicinal
chemistry. Late-stage functionalization (LSF), in this regard, provides the opportunity to
directly and economically modify both early building blocks or advanced molecules
which circumvents the necessity for establishing alternate synthetic routes to install
new substitution patterns. Synthetic methods for selective activation and modification of
C–H bonds on organic scaffolds are available for both directed and non-directed, as
well as chemo- and site-selective transformations. In addition to providing control over
the modification sites, the abundance of LSF procedures enables accessing a variety of
different modifications. To date, however, only few applications of LSF in drug
discovery have been published due to the difficulty in predicting the reaction outcome if
multiple functional groups and various types of C–H bonds with different bond strength,
electronic properties and steric environments are present. As a consequence, running
a successful LSF campaign using traditional workflows often requires time-consuming
resource-intensive experimentation, which might not feasible during medicinal
chemistry explorations. To approach this challenge, we have built a high throughput
experimentation (HTE) platform that provides semi-automated miniaturized low-volume
screenings to accelerate the performance of multiple transformations in parallel. In a
proof-of-concept study, we have investigated the regioselective introduction of boronic
esters into advanced drug-like molecules, which serve as synthetic handles for
installing a variety of different functional groups. As starting materials for our HTE
approach, we have chosen 23 structurally diverse drug molecules, 12 drug-like
fragments and 5 frequently occurring literature substrates, which culminated in the
performance of 956 reactions. Together with the group of Gisbert Schneider, we have
used the generated reaction data along with a literature dataset to train graph neural
networks. Our computational model, which considered the influence of steric and
quantum mechanical information, correctly predicted the reactivity of 81% of novel
substrates, while reaction yields for diverse reaction conditions were predicted with a
mean absolute error margin of 4–5%. The regioselectivity of the major products was
accurately captured in up to 90% of the cases studied.

References:
[1] D. F. Nippa, et al., CHIMIA 2022, 76, 258.
[2] D. F. Nippa, et al., Nat. Chem. 2024, 16, 239–248.
[2] D. F. Nippa, et al., Commun. Chem. 2023, 6, 256.
[2] D. F. Nippa, et al., ChemRxiv 2023, doi:10.26434/chemrxiv-2023-nfq7h-v2.

On 6 June 2024, Prof. Dr Stefan Laufer (University of Tübingen) will give a guest lecture entitled ‘Academic Drug Discovery: Fiction - Facts - Fantasy?’. The lecture will begin at 7 pm in the HIPS (Building E8 1, seminar room ground floor).

Translational drug discovery in an academic environment was long considered impossible, but is now a reality at some locations, including in Germany. The ‘Tübingen Center for Academic Drug Discovery’ is a platform within the Excellence Strategy of the Eberhard Karls University of Tübingen and is based on the chemical/pharmacological validation of genetically identified new drug targets, using the example of a siRNA-based screen on liver cells to identify two new targets that could be brought to proof of concept in humans purely academically or, in the other case, as part of a VC-based spin-off. The lecture will present these two case studies as examples.

On 5 June 2024, Elodie Laine (Laboratory of Computational and Quantitative Biology, IBPS, Sorbonne University) will be our guest from 3 pm and will speak on the topic ‘Expert-guided protein language models enable accurate and blazingly fast fitness prediction’. The event will take place in seminar room 0.01 in building E2 1.

Exhaustive experimental annotation of the effect of all known protein variants remains daunting and expensive, stressing the need for scalable effect predictions. I will introduce VespaG, a blazingly fast single amino acid variant effect predictor, leveraging embeddings of protein Language Models as input to a minimal deep learning model. To overcome the sparsity of experimental training data, we created a dataset of 39 million single amino acid variants from the human proteome applying the multiple sequence alignment-based effect predictor GEMME as a pseudo standard-of-truth. Assessed against the ProteinGym Substitution Benchmark (217 multiplex assays of variant effect with 2.5 million variants), VespaG achieved a mean Spearman correlation of 0.48 +/- 0.01, matching state-of-the-art methods such as GEMME, TranceptEVE, PoET, AlphaMissense, and VESPA. VespaG reached its top-level performance several orders of magnitude faster, predicting all mutational landscapes of the human proteome in 30 minutes on a consumer laptop (12-core CPU, 16 GB RAM).

On 5 June 2024, Alexey Gurevich, Helmholtz Institute for Pharmaceutical Research Saarland (HIPS), will speak about ‘Hunting new antibiotics with computer science’. The lecture will take place at 12:15 pm in CISPA, Building E1 5, Room 002, Saarbrücken Campus.

WHO describes antimicrobial resistance (AMR) as one of the top global public health and development threats, while the World Bank estimates its additional healthcare burden as US$ 1 trillion by 2050. One strategy to combat AMR is significantly speeding up and reducing the cost of discovering new antibiotics, particularly those existing in nature but evading all currentparticularly attempts to find them. Surprisingly, computer scientists play an increasingly important role in this endeavor. Modern biotechnology allows us to amass vast amounts of data on natural antibiotics and their tiny producers. However, specialized software and algorithms are the key to unlocking this wealth of information and turning it into medically important discoveries.

In my talk, I will demonstrate how computer science methods, from classical graph and string algorithms to deep learning, help transform antibiotic discovery into a high-throughput technology and realize the promise of already collected and rapidly growing biological datasets. The particular focus will be on analyzing sequencing (strings over the DNA alphabet of {A, C, G, T} letters) and mass spectrometry data (two-dimensional arrays of floating point values). In both cases, we deal with noisy experimental data, often rely on heuristics to make the processing time reasonable, and finally, get biologically relevant findings suitable for verification by our wet-lab collaborators.

Presenter: Alexey Gurevich

On May 23, 2024, Alexander Tkatchenko (University of Luxembourg) will talk about ‘AI (R)Evolution in (Quantum) Chemistry and Physics’. The guest lecture will take place at 16:30 in building E2.1, room 0.01 (Saarbrücken campus).

Abstract

Learning from data has led to paradigm shifts in a multitude of disciplines, including web, text and image search and generation, speech recognition, as well as bioinformatics. Can machine learning enable similar breakthroughs in understanding (quantum) dynamics of molecules and materials? This is an especially relevant question given that quantum mechanical predictions are now within experimental uncertainties for complex molecular systems (https://www.nature.com/articles/s41586-023-06587-3).

Aiming towards a unified machine learning (ML) model of molecular interactions in chemical space, I will discuss the potential and challenges for using ML techniques in chemistry and physics. ML methods can not only accurately estimate molecular properties of large datasets, but they can also lead to new insights into chemical similarity, aromaticity, reactivity, and molecular dynamics. For example, the combination of reliable molecular data with ML methods has enabled a fully quantitative simulation of protein dynamics in water (https://www.science.org/doi/full/10.1126/sciadv.adn4397). While the potential of machine learning for revealing insights into molecules and materials is high, I will conclude my talk by discussing the many remaining avenues for development.

On 15 May 2024, Prof. Dr. Matthias Rarey (Computational Molecular Design, Hamburg University) will speak on ‘Algorithms that Matter: Examples from Cheminformatics and Structure-Based Drug Design’ as part of the nextTALK series.

The lecture will take place from 9 to 10 a.m. in building E2 1, room 0.01 (Saarbrücken campus).

On May 8, 2024, Rayan Chikhi (Institut Pasteur) will give a guest lecture on "Sequence bioinformatics on really big data: petabase-scale sequence alignment catalyses viral discovery". The lecture starts at 3 pm in building E2.1, room 0.01 (Saarbrücken campus).

Abstract: Petabytes of valuable sequencing data reside in public repositories, doubling in size every two years. They contain a wealth of genetic information about viruses that would help us monitor spillovers and anticipate future pandemics. We have developed a bioinformatics cloud infrastructure, named Serratus, to perform petabase-scale sequence alignment. With it we analyzed all available RNA-seq samples (5.7 million samples, 10 petabytes) and discovered ten times more RNA viruses than previously known, including a new family of coronaviruses (Edgar et al, Nature, 2022). In this talk, I will present the computational infrastructure and the biological analyses. More recently, we have performed SRA-scale genome assembly, and I will briefly talk about the current state of this project (Logan/IndexThePlanet).

On April 8, 2024, Kenneth Atz, AI scientist at Roche, will give a talk on "Medicinal Chemistry in the Era of Geometric Deep Learning: From Generative Design to Synthesis Prediction" which will be the first of our nextTALK lecture series. The talk will be held in person at HIPS (building E8.1, seminar room, ground floor) at 3 pm.