About the PULSE program

This area integrates the disciplines of genetics and biochemistry with the ambition to fully understand how cells, the fundamental unit of life, function.  Progress is made by analyzing the molecular components in time and space, from single molecules to native tissue environments at high resolution. This research area is rapidly evolving with the development and application of novel high-density data driven methods, often relying on machine learning, artificial intelligence, or other computational techniques to analyze, integrate and interpret large sets of imaging data. The ultimate goal is to create predictive models for core functions of cells and tissues.

At SciLifeLab, research in this area is aligned with the EU definition of precision medicine, ie, “characterization of individuals’ phenotypes and genotypes to tailor therapeutic strategies, assess disease predisposition, and enable timely, targeted prevention.” Accordingly, here the aim is to integrate molecular and clinical data to identify biomarkers for disease risk and diagnosis, evaluate drug responses, and support health monitoring. Research projects are expected to contribute with strong capabilities inmachine learning, AI and other computational tools. Possibilities exist to mine strong assets in Sweden, such as molecular data (e.g. omics), imaging, electronic healthcarerecords, longitudinal patient and population registries, biobanks and digital monitoring data.

This subject area concerns research using large experimental, clinical, or pathogen surveillance data in innovative ways to transform our understanding of pathogens, their interactions with hosts and the environment, and how they are transmitted through populations. The aim is to meticulously analyze pathogen-host model systems, exploiting multidimensional and genome-scale experimental data that are now routinely attainable, and extending to population-scale genetic, clinical, or public health data from pathogen surveillance efforts and biobanks. The ultimate goal is to enhance the ability to anticipate, prevent, and respond to infectious disease outbreaks, ensuring a proactive approach to safeguarding public health.

Projects within this area address life from a trans-disciplinary perspective, encompassing molecules and cells to ecosystems. Researchers are leveraging the substantial data streams generatedby high-throughput sequencing of genomes and biomes,continuous recording of video and audio in the wild, high-throughput imaging of biological specimens, and large-scale remote monitoring of organisms and habitats. Novel insights are likely to emerge by exploiting the development and applicationof novel methods relying on machine learning, artificial intelligence, or other computational techniques, enhancing our understanding of population genetics, adaptation, speciation, convergent evolution, and functional constraints. A deeper understanding of the complex interactions within ecosystems and their implications for planetary health is needed to find solutions to impede the ongoing loss of biodiversity.

This is an exceptional opportunity to conduct research at the intersection of artificial intelligence (AI)/Machine Learning (ML), data science, and drug discovery. The fellow will lead the design, development, and implementation of novel AI/ML methods for preclinical drug discovery of small molecules, proteins, or oligonucleotides, focusing on developing and applying AI/ML methods in hit-to-lead optimization projects for any of these drug modalities.

Key project areas include, but are not limited to, the development of AI/ML tools, web-based workflows, and novel approaches in deep learning, de novo design and generative AI for drug discovery. Candidates are encouraged to apply AI/ML methodologies across preclinical drug discovery project based on small molecules, proteins, oligonucleotides, or emerging “new modalities”. Examples of areas of research could be: novel molecular representations and generation, chemical space navigation, ultra-large virtual screening, multi-parameter optimization, ligandability assessment, protein design, RNA and oligonucleotide structure prediction, prediction of on/off-target oligonucleotide binding and effect, computational characterization of enzymatic RNA cleavage, antibody developability, and de novo design. We also welcome proposals from emerging fields in drug discovery, including explainable AI, and integrating ‘first-principles’ with AI/ML (e.g., quantum machine learning approaches).

Examples of challenges in this area:

• There is very limited access to high-quality, standardized datasets, combined with imbalanced data (e.g., active vs. inactive compounds) which hampers model training and generalizability. How to access these data sets?
• Many models, particularly deep learning-based ones, suffer from a lack of interpretability, making it difficult to validate predictions and gain insights into molecular mechanisms. How can new methods be developed to solve this problem?
• Scalability is another barrier, as handling large datasets or running complex simulations requires significant computational resources. How to speed up things? 
• The interdisciplinary knowledge gap between AI researchers and domain experts in chemistry and biology slows progress, as effective collaboration requires bridging expertise across these fields. How could these barriers be torned down?
• Balancing stability, permeability, and bioactivity in new modalities such as peptides, targeted protein degraders, and oligonucleotides is challenging due to conflicting design strategies. Therefore, developing AI/ML-driven design principles with improved pharmacokinetic properties is crucial for modulating ‘undruggable’ targets. How could this be achieved?

The successful candidate will join a vibrant, multidisciplinary research environment, collaborating with bioinformaticians, structural biologists, chemists, and drug discovery experts.

Contact: vasanthanathan.poongavanam@scilifelab.uu.se

Through the OligoNova Hub units in Gothenburg, SciLifeLab-DDD offers state-of-the-art capabilities for oligonucleotide-based drug discovery, including but not limited to antisense and siRNA-based therapeutic modalities, as well as modalities that conjugate oligonucleotides to SciLifeLab-DDD’s other modalities such as proteins and small molecular ligands. In-house capacity for bioinformatics and AI/ML-guided oligo design, oligonucleotide synthesis, automated and high-throughput analyses of cellular effects through qPCR, NGS, mass spectrometry, imaging, and other techniques is complemented with advanced proteomics, NMR, and additional imaging modalities including Nano-SIMS within the local core facilities organization and other collaborators.

New technologies to design and analyse therapeutic oligonucleotides to support the development of optimally efficacious and safe therapeutic agents are continuously developed and applied within the focus area. Current examples include the development of a modular reporter assay system to evaluate target knockdown, new computational techniques for nucleic acid structure prediction, and biophysical technologies for the characterization of oligonucleotide binding interactions.

Of great interest are new technologies and approaches that could streamline any aspect of the discovery process, provide new molecular and/or mechanistic insight into the action and safety of therapeutic oligonucleotides, provide new means of measuring oligonucleotide distribution, exposure, and effects in vitro and in vivo, or enable the design of novel tissue- and cell type-selective oligonucleotide delivery strategies. In addition to experimentally oriented projects, AI/ML-approaches or other computational techniques could be integrated project components.

Examples of challenges in this area:

• Our ability to identify the best candidates for in vivo testing of therapeutic oligonucleotides are limited. How can new methods be developed to characterize and prioritize therapeutic oligonucleotides in large libraries with regard to efficacy and/or safety?
• Intracellular localization is key for functional ASOs and siRNA therapeutics. How can new methods be developed to measure tissue- and (sub)cellular distribution of oligonucleotide therapeutics?
• Today’s therapeutic oligonucleotides are essentially restricted to diseases in the liver, the central nervous system and the eye. How can novel strategies be developed for enhancing delivery of oligonucleotide therapeutics to other tissues and cell types through conjugation to targeting molecules, and new methodologies to identify and characterize such targeting strategies

Contact: par.matsson@gu.se

Within the SciLifeLab Display and Selection Technologies focus area we have access to state-of-the-art platforms for discovery of therapeutic antibodies, small molecules and oligonucleotides. These include e.g. phage display libraries as source of human antibodies, DNA-encoded chemical libraries (DELs) for small molecules, automated systems to facilitate high-throughput screening and instruments to study binding of therapeutic candidates to proteins and cells. Through close collaborations with other SciLifeLab units, local university networks and CROs additional instruments can be accessed e.g. mass spectrometry, DNA sequencing (Sanger, NGS, long-read) and various imaging techniques.

Technology development is continuously ongoing within the focus area. Currently a single-domain antibody library in phage display format as well as new DELs are under construction, antibody tools for targeted protein degradation and intracellular delivery are developed, new reporter systems for screening of oligonucleotides are set up and long-read sequencing is explored as a complement to traditional antibody screening.

We welcome technology development proposals improving any step of the early drug discovery process. Areas of interest could include e.g. novel selection and screening strategies, alternative display technologies, new therapeutic antibody formats, strategies for optimization of binding affinity and strategies for improved developability. AI-driven approaches and NGS/long-read sequencing could be integrated elements in such development work.

Examples of challenges in this area:

• Presently, recombinant protein displayed on beads are used for DEL screening. Can a technology be developed that allows for in situ selection in human cells or in bacteria using large DELs?
• NGS analysis captures the full diversity of antibody pools isolated from display libraries in contrast to traditional screening which merely scratches the surface. How to bypass screening and select antibodies based on NGS data?
• Preclinical development of antibody drug candidates often includes resource-demanding engineering to improve e.g. affinity or various developability aspects.  Can new strategies be developed to make optimization more efficacious?
• Novel technologies and vectors that generate a better therapeutic window for CAR applications are in demand. How can this be accomplished?
• There is a need for new display/array/antibody/peptide formats, or combos of such technologies, to facilitate isolation of multi-functional drugs such as PIAs. How can such technologies be developed?

Contact: mikael.mattsson@immun.lth.se

Proximity-Inducing Agents are an emerging approach of therapeutics that leverage the ability to bring two or more biomolecules into proximity to induce specific biological effects. These agents are based on technologies such as bi-specific antibodies, antibody-drug conjugates, or various small molecule approaches for intracellular re-wiring of protein localization. Emerging technologies for Targeted Protein Degradation (TPD), such as PROTACs (Proteolysis Targeting Chimeras) and AbTACs (antibody-based targeted protein degradation agents), show promise in precisely targeting and degrading disease-causing proteins, potentially offering new treatment avenues that conventional drugs may not address.

We offer exceptional opportunities to conduct research that will shape the technological advancement of proximity inducing therapeutics based on modalities that is compatible with areas within scope for the SciLifeLab DDD platform, i.e. small molecules, proteins, oligonucleotides, or new modalities. Cell-based therapies are out-of-scope.

Depending on the candidate’s background and interests we support proposals related, but not limited, to: small molecule degraders, antibody-based degraders for extra-cellular or membrane proteins, targeted delivery of payloads (oligonucleotides, small molecules, radioligands, degraders), bi-specific antibodies, etc. Currently, we have toolboxes, assays, and methodology for small molecule- or antibody-based degraders and targeted delivery available. We offer industry standard facilities and infrastructures for e.g. organic synthesis – including oligonucleotides, biochemical and cellular assays, biophysics, ADME, protein expression, phage-display selections, computational and machine learning capabilities, etc. SciLifeLab offers opportunities to access additional world-class infrastructures.

Examples of challenges in this area:

• Proximity-inducing agents is a powerful concept to invent new therapeutic approaches for pharmaceutics. How can we identify and validate new effector-target pairs to broaden the scope of proximity-inducing agents (PIAs) beyond protein degradation? 
• DNA-encoded library screening in the presence or absence of presenter proteins has been used to rationally identify PIAs. Which alternative methods could be explored to identify PIAs (small molecules, peptides, antibodies, etc.)?
• We have toolboxes of E3-binders – both small molecules and antibody fragments. How should we utilize these platforms for targeted protein degradation in underexplored areas of biology to drug the undruggable?

Contact: Per.Arvidsson@scilifelab.se

Challenges in drug discovery span the entire process, from initial target identification and lengthy, expensive preclinical and clinical trials to regulatory hurdles and patient-specific factors. Key issues include a lack of understanding of disease biology, failures in animal models, the high cost and failure rates of drug development, poor patient recruitment, the need for personalized medicine due to patient heterogeneity, and the lack of validated biomarkers. The SciLifeLab Drug Discovery and Development platform (DDD) performs target-based drug discovery and the main focus is on the early stages of Drug Discovery, i.e. hit identification, lead generation and first preclinical proof of concept studies of prototype drugs in models with translational value for future clinical studies.  

In early drug discovery, challenges include target identification, such as accurately identifying and validating a disease-relevant therapeutic approach and understanding how that approach may interfere in the disease pathway; hit identification, involving the mining of gigantic libraries of putative drugs (small molecules, peptides, proteins e.g antibodies and nucleic acids) that interact with the target and that are suitable as starting points for further drug development; and lead generation, which requires improving properties of these initial “hits” into more potent, selective, and drug-like molecules while minimizing off-target effects and potential toxicity. Overcoming these challenges requires integrating diverse data, improving computational tools, understanding disease complexity, and increasing the efficiency and reliability of experimental processes. 

Examples of challenges in early drug discovery

Accuracy and Validation:

Identifying a therapeutic approach for a defined target that is truly involved in the disease pathway and is “druggable” (meaning a drug can effectively interact with it) is difficult. New methods to rationalize how novel (previously “undruggable”) targets could be explored using new therapeutic modalities capable of inducing novel pharmacology, e.g. proximity induction for cellular re-trafficking, targeted degradation, etc. A challenge is also to ensure the identified target’s role is fully understood, including both on-target effects and potential off-target effects that could lead to side effects. 

Data Heterogeneity in AI:

Identifying true hits from massive libraries is often difficult due to the immense number of molecules and the potential for false positives or negatives. While Artificial Intelligence offers solutions, the availability of high-quality, diverse datasets needed to train AI models is a significant hurdle. 

Predicting Efficacy and Toxicity:

A key challenge is ensuring that the lead candidate does not interact with other unintended targets in the body, which can lead to toxicity and side effects. 

Accurately predicting how a lead compound will behave in a living organism, particularly concerning both efficacy and potential toxicity, is a major challenge. 

Bridging In Vitro to In Vivo:

Animal models often fail to accurately replicate human diseases, leading to failures when drugs are tested in humans. Hence, successfully translating promising in vitro (laboratory) results into in vivo (animal or human) efficacy remains a significant hurdle, as biological systems are complex and can lead to different outcomes.

Lack of Validated Biomarkers:

A shortage of reliable diagnostic, affordable and therapeutic biomarkers makes it harder to identify the right patients for trials and monitor drug effectiveness.

DDD has focused technology development efforts into four areas of importance for future pharma and biotech therapeutics. Machine learning has the potential to make the drug discovery process more cost effective and will be applied everywhere in drug discovery. How can we adapt these technologies in the best way at DDD? Display and selection technologies have the potential to increase our ability to mine gigantic libraries to identify the most relevant starting points for development of prototype drugs. Which technologies are best suited for DDD? Proximity inducing agents has the potential to pave the way for polypharmacology, i.e. the drug discovery approach where a single molecule is designed to simultaneously act on multiple biological targets to achieve a novel and more impactful therapeutic outcome. Therapeutic oligonucleotides have the potential as personalized therapies with extended treatment effects that may overcome challenges with e.g. “druggability” and compliance. However, targeted delivery to cells and organs, with the exception of liver and CNS, is problematic.

Follow links for further reading and more detailed information on challenges in drug discovery of new modalities for therapeutic interventions:  Emmerich 2021 ; Blanco 2020