RNA does its work in three dimensions, so we start from the shape. The Laboratory of RNA Design and Therapeutics combines deep learning and experimental methods to predict those shapes and to design aptamers, therapeutic mRNA and chemically modified RNA.
RNA does its work in three dimensions. We predict that geometry with machine learning, then use it to design molecules that bind, fold and function the way we intend.
End-to-end models that go straight from sequence and alignment to atomic coordinates, built on the OpenFold architecture and trained on curated structural data.
Which pockets does a folded RNA actually present, and what binds them? We combine structure prediction with selection experiments to make binding predictable.
Chemical modification changes both folding and function. We model pseudouridine (Ψ) thermodynamics and apply it to translation control elements.
Models are only as good as the data behind them. We curate structural RNA data and keep train/test splits honest, family by family.
Computation and experiment are one cycle here, not two groups. A model proposes a structure or a binder, the lab measures whether it behaves as predicted, and the measurement goes back into the training data.
Structure prediction and ligand-conditioned generation produce candidate folds, aptamer sequences and modified constructs, ranked by the model's own confidence.
Candidates go into SELEX, binding assays and SHAPE-MaP probing. Predicted structure and predicted affinity are compared against what the molecule actually does.
Hits and failures alike become training and benchmark data, so the next round of predictions is wrong in fewer places than the last.
predict → test → feed back → predict · every dataset we generate is released with the model that learned from it
Everything the lab produces is released openly, with permissive licensing, so results can be reproduced and reused.
End-to-end deep learning approach for RNA 3D structure prediction, built on OpenFold. Apache-2.0.
ActiveAutomated web pipeline that searches for RNA homologs, aligns them and assesses the secondary structure of the resulting alignment. Published in Nucleic Acids Research (2025).
Structurally non-redundant dataset of RNA structures with sequence-identity-aware splits for training and evaluation.
ActiveToolbox for handling RNA 3D structures: format conversion, renumbering, cleaning, batch analysis.
New results, people joining, talks and preprints.
Hands-on SELEX training at CARD, hosted in Günter Mayer’s group at the LIMES Institute, University of Bonn. CARD runs an automated platform for generating customized aptamers and makes it available to academic and industrial researchers as a non-profit service, together with Marcus Menger’s Functional Nucleic Acid – Aptamer laboratory at the Fraunhofer Institute for Cell Therapy and Immunology in Potsdam. The visit brings the experimental side of aptamer selection into the lab, next to the models we build for it.
“Usually once you become a PI, you stop pipetting. I’m doing the reverse PI-ing!” — Magnus, at the bench in Bonn.
Hybrid invited lecture at the Institute of Structural Biology, University Hospital Bonn, on 18 September 2026. Hosted by Radosław Nowak.
Jadwiga starts this month, taking on in vitro selection of RNA aptamers and testing computationally designed binders at the bench — the step that closes the loop between predicted structures and measured affinity. Funded by the NCN SONATA BIS-15 grant. Welcome!
Together with Juliusz and Grzegorz Łach we spent a day with the seventeen participants of the Collegium Invisibile Summer School, run with Fundusz Zdolni (formerly the Polish Children's Fund) at the rectors' palace in Ciążeń. The students are sixteen to eighteen years old and heading in every direction — medicine, engineering, linguistics — so we talked about what computational RNA biology actually looks like from the inside.
“OpenRNAFold: an end-to-end approach for RNA 3D structure prediction” — talk given at the BIT26 conference, 18–20 June 2026.
Marcin gave the last Computational Biology and Bioinformatics seminar before the summer break, on 10 June in room 3250 at Banacha 2 — an end-to-end, AlphaFold-style approach to RNA 3D structure prediction, built on carefully curated training data (RNA3DB) and high-quality alignments (RNAhub) to avoid the inflated accuracy that comes from overlap between training and test sets. Hosted by Aleksander Jankowski on behalf of the BOB organisers.
The Laboratory of RNA Design and Therapeutics starts at the Centre of New Technologies, following a postdoc in Elena Rivas’s group at Harvard.
Five-year funding to build the group around deep learning for RNA structure and RNA–small molecule interactions.
A group of computational and experimental RNA scientists — and, right now, a few desks still waiting for the right people.
Computational RNA biologist. PhD with Janusz Bujnicki at IIMCB Warsaw, experimental training with Magda Konarska, an internship with Rhiju Das at Stanford, and a postdoc with Elena Rivas at Harvard. Works on RNA 3D structure prediction, RNA datasets and open-source tooling for structural RNA biology.
Runs in vitro selection of RNA aptamers and tests computationally designed binders at the bench, closing the loop between predicted structures and measured affinity. Funded by the NCN SONATA BIS-15 grant.
Generative models that build RNA structure: ligand-conditioned diffusion, denoising of backbones and bases, and the evaluation and data pipelines that keep the benchmarks honest. Funded by the NCN SONATA BIS-15 grant; the call is to be opened in 2026.
A new position for someone with a language-model background. RNA sequence, alignments and structure all lend themselves to LLM-style architectures, and this project is about building and evaluating them for RNA design and structure prediction.
A four-year computational doctoral project on RNA design: generative and language-model approaches to ligand-binding RNA, benchmarked against the structures and selection data the lab produces. Funded by the NCN SONATA BIS-15 grant, through the UW doctoral school.
The lab's methods are built and tested together with groups in comparative RNA genomics and sequence analysis.
Harvard University
Comparative RNA genomics and covariation analysis. Joint work on OpenRNAFold and on evidence for conserved RNA structure in alignments, continuing from Marcin's postdoc in the Rivas lab.
University of Western Australia
First author of RNA3DB and the driving force behind its structurally non-redundant splits. Joint work on dataset curation and on benchmarks that keep RNA structure prediction honest.
Replace the entries below with your selected papers — title, author list, journal and DOI.
Manuscript in preparation
Nature Methods 22(2), 399–411 · DOI: 10.1038/s41592-024-02543-9
Nucleic Acids Research 53(W1), W496–W502 · DOI: 10.1093/nar/gkaf342
Journal of Molecular Biology 436(17), 168552 · DOI: 10.1016/j.jmb.2024.168552
A master's programme built jointly by the University of Warsaw and the International Institute of Molecular and Cell Biology in Warsaw, for students heading into modern biological and biomedical science. The curriculum runs from the design of nucleic acid-based drugs and their molecular mechanisms through delivery in lipid nanoparticles to preclinical studies and commercialisation — close enough to what this lab does that students from the programme are welcome to ask about projects with us.
Ten years of RNA Club. It started in October 2016 as an RNA-focused spin-off of Do Science!, with the support of prof. Jacek Kuźnicki at IIMCB, and soon afterwards secured RNA Salon funding from the RNA Society. Since then it has held keynote evenings with Rhiju Das, Eric Westhof and Matthew Disney, run joint workshops with NGSchool, and hosted a long series of members' meetings, mostly at IIMCB — a local meeting for everyone in Warsaw who works on RNA, across institutes, from computation to experimental work.
A Special Interest Group on nucleic acid 3D structure prediction — experimental data, training and test sets, RNA alignments, ligand interactions, conformational ensembles. The community meets online every two weeks, and the talks are recorded and published openly. How the CASP Special Interest Groups work, this one included, is described in a community paper we co-authored in Proteins (2025).
A science discussion club in Warsaw — lectures and informal meetings where researchers talked about their work and about science itself. It began in April 2013 with a meeting with Robert Huber, took its name that October, and over nine years hosted three Nobel laureates (Robert Huber, Brian Kobilka, Venki Ramakrishnan) alongside many other speakers. Institute-independent and open to everyone, it was run by PhD students and postdocs from IIMCB and Biocentrum Ochota, with support from EMBO, IIMCB and Biocentrum Ochota, and it also produced SciEvents, a shared calendar of bioscience events in Warsaw. doscience.iimcb.gov.pl
We look for people at every level — bachelor, master, PhD, postdoc — who are curious about RNA structure and comfortable moving between code and experimental work. Experience in machine learning, structural biology or experimental techniques is all welcome.
University of Warsaw, Ochota campus.
