Research | Pacesa Lab

Our Research

We are especially interested in studying biomolecular systems that are heterogeneous, state-dependent, mechanistically rich, and inconveniently real. Biology is messy and encompasses a myriad of first principles that we need to first understand to be able to reconstruct it, so our research focus is wide. We love ambitious, quirky, and niche projects and applications. The more unusual and out of distribution the problem is, the better we can probe the limits of existing approaches. In the near term, we are focusing on the study of protein-nucleic acid interactions, as they are dynamic, structurally versatile, and central to life. Thematically, the lab is currently divided into three complementary development areas:

◈

Increasing the throughput and real world modelling accuracy of structural biology methods

Structural biology is technically demanding, slow, and largely limited to static snapshots. Yet it is the primary source of data for understanding and modelling biomolecules. We are automating and scaling structure determination to generate the rich datasets needed to train better predictive models, with the long-term goal of simulating entire cells, tissues, and organisms under defined conditions.

◍

Developing design methods with the philosophy:"one click, one working design"

Computational design pipelines are often difficult to run for non-specialists and require so much wetlab screening that they become projects of their own. We want to democratise biomolecular design and improve it to a point where it becomes as trivial as clicking a button. Therefore, we view the real-world experimental performance as the only benchmark that matters, not just showing nice in silico scores.

◌

Unlocking the power of de novo designed proteins in research and therapy

De novo designed proteins are not constrained by evolution and can have properties that classical reagents and antibodies simply cannot match. This enables applications in areas where other types of modalities fail. We therefore assess their safety, immunogenicity, and developability profiles in complex environments and push to translate them into real world applicable reagents and therapies.

Our aim is to understand biomolecular systems with enough precision to design and deploy them deliberately: ranging from a protein binding interface engineered atom by atom, to molecular machines operating inside living cells, to synthetic organisms built from first principles. While that goal may be distant, every method, dataset, and insight we develop is oriented towards eventually achieving it.

Increasing the throughput of cryoEM and X-ray crystallography

▸

We develop high throughput experimental sample preparation approachces for cryoEM and X-ray crystallography

▸

This is accompanied by development of novel and automated data processing pipelines for fast screening and accurate model building

▸

Our goal is to diversify and increase the size of current structural databases used for deep learning model training

Incorporating heterogeneity into structural models of biomolecules

▸

Biology is inherently heterogeneus, especially on scales relevant to life and therefore we need to change our models to account for it

▸

We model the real heterogeneity of biomolecules, without relying on simulations with arbitrary priors

De novo designed binder engaging a therapeutic target

Designing effectors engaging complex biomolecular substrates

Programmable biomolecular design

▸

We build user-friendly tools that make advanced computational design accessible to everyday biologists rather than only experts

▸

We are pushing from protein-protein interactions today into protein–nucleic acid interaction design as an emerging frontier

▸

We aim to design complex molecular machines and networks that eventually will build up to de novo cellular systems and organisms

Translating de novo proteins towards real life applications

▸

Every design is validated in the wetlab using purified components, biophysical characterization, and tested in cellular and translational assays

▸

We aim to demonstrate the safety of de novo designed proteins as a new class of efficient human therapeutics

▸

We also apply their favourable properties in other niche areas outside of clinical applications

Why focus on protein-nucleic acid systems?

Protein-nucleic acid interactions are especially fascinating because they are highly dynamic, structurally diverse, and central to biology. Nucleic acids are often imagined to be simple linear molecules with rigid pairing rules, while in fact they are anything but that and real biology breaks those rules constantly. That versatility makes them difficult to model and therefore unusually informative to study.

It is the second-best described biomolecule after proteins from a structural and biophysical point of view.

They are dynamic, compositionally diverse, and capable of much richer behavior than standard textbook paradigms suggest.

If we can design molecules capable of interfacing with nucleic acids in a defined manner, we gain leverage over genomes, cellular state, and eventually more general programmable biology.

Decoding the principles underlying biomolecular structure and the dynamics of large biological assemblies across scales