Career and Education
Postdoctoral Scholar and Senior Scientist
Since 2021
ETH Zurich, Switzerland
After my PhD, I transitioned from experimental to computational chemistry specializing in geometric deep learning. In the group of
Sereina Riniker, I explored deep neural networks
to replace expensive quantum mechanical calculations in molecular dynamics simulations. Together with my colleague
Moritz Thürlemann, we
developed and applied anisotropic message passing (AMP)
to encode quantum mechanical symmetries in neural network potentials, resulting in equivariant message passing
graph neural networks.
Doctor of Sciences
2021
ETH Zurich, Switzerland
My PhD thesis with
Erick M. Carreira focused on the synthesis and biological
investigation of mutanobactin D. We could show that this molecule is a key signaling molecule within
the human microbiome and responsible for maintaining homeostatis of the human host and microbial colonizers.
Toward the end of my PhD, I became more interested in computational methods within chemistry, especially
quantum mechanis and deep neural networks.
Master of Science
2016
University of Munich, Germany
BSc. and MSc. degrees in chemistry and biochemistry at the University of Munich.
I worked with Dirk Trauner on the total synthesis of natural products and wrote my thesis
at the University of Toronto with Mark Lautens
in transition metal catalysis. I also interned at the medicinal chemistry department of Bayer.
Scientific Publications
- Pultar, F.;† Katzberger, P.;† Riniker, S. Transferring Knowledge from MM to QM for a Graph Neural Network Based Implicit QM Solvent.
Preprint: ChemRxiv (2025) - Pultar, F.;† Brandl, P.;† Lüthy, L.;† Hansen, M. E.; Jemini, N.; Faris, J. H.; Yasmin, S. O.; Kamenik, A. S.; Arnold, R.; Ebert, M.-O.; Wolfrum, S.; Lokey, R. S.; Riniker, S.; Carreira, E. M. Mutanobactin D from the Human Microbiome: Chemistry, Biology, and Molecular Dynamics Studies.
Preprint: ChemRxiv (2025) - Pultar, F.;† Thürlemann, M.;† Gordiy, I.; Doloszeski, E.; Riniker, S. Neural Network Potential with Multiresolution Approach Enables Accurate Prediction of Reaction Free Energies in Solution. J. Am. Chem. Soc. 2025, 147, 6835–6856.
Preprint: arXiv: 2411.19728 (2024).
DOI: 10.1021/jacs.4c17015 - Nikolic, S.; Alastra, G.; Pultar, F.; Lüthy, L.; Stadlinger, B.; Carreira, E. M.; Bugueno, I. M.; Mitsiadis, T. A. Mutanobactin-D, a Streptococcus mutans Non-Ribosomal Cyclic Lipopeptide, Induces Osteogenic/Odontogenic Differentiation of Human Dental Pulp Stem Cells and Human Bone Marrow Stem Cells. Int. J. Mol. Sci. 2025, 26, 1441.
DOI: 10.3390/ijms26031144 - Poliak, P.; Bleiziffer, P.; Pultar, F.; Riniker, S.; Oostenbrink, C. A Robust and Versatile QM/MM Interface for Molecular Dynamics in GROMOS. J. Comp. Chem., 2025, 46, e70053.
DOI: 10.1002/jcc.70053 - Cotter, E.; Pultar, F.; Riniker, S.; Altmann, K.-H. Experimental and Theoretical Studies on the Reactions of Aliphatic Imines with Isocyanates. Chem. Eur. J. 2024, 30, e202304272.
Preprint: ChemRxiv (2023).
DOI: 10.1002/chem.202304272 - Pultar, F. Total Synthesis and Biological Evaluation of Mutanobactin D from the Human Microbiome. Ph.D. Dissertation, ETH Zurich, 2021.
DOI: 10.3929/ethz-b-000505539 - Pultar, F.; Hansen, M. E.; Wolfrum, S.; Böselt, L.; Fróis-Martins, R.; Bloch, S.; Kravina, A. G.; Pehlivanoglu, D.; Schäffer, C.; LeibundGut-Landmann, S.; Riniker, S.; Carreira, E. M. Mutanobactin D from the Human Microbiome: Total Synthesis, Configurational Assignment, and Biological Evaluation. J. Am. Chem. Soc. 2021, 143, 10389–10402.
DOI: 10.1021/jacs.1c04825 - Pultar, F.;† Johnson, T.;†; Menke, F.; Lautens, M. Palladium-Catalyzed α-Arylation of Vinylogous Esters for the Synthesis of γ,γ-Disubstituted Cyclohexenones. Org. Lett. 2016, 18, 6488–6491.
DOI: 10.1021/acs.orglett.6b03394 - Tambornino, F.; Sappl, J.; Pultar, F.; Cong, T. M.; Hübner, S.; Giftthaler, T.; Hoch, C. Electrocrystallization: A Synthetic Method for Intermetallic Phases with Polar Metal–Metal Bonding. Inorg. Chem. 2016, 55, 11551–11559.
DOI: 10.1021/acs.inorgchem.6b02068
Teaching
- To bridge the gap between experimental and computational chemistry, I developed and taught together with Eno Paenurk and Patrick Finkelstein a online course Introduction to Computational Chemistry at ETH Zurich and Yale University. The course was well attended by undergraduate and graduate students as well as faculty members.
- While at ETH Zurich, I was a teaching assistant for the courses Algorithms and Programming in Chemistry, Introduction to Asymmetric Synthesis, and Methods and Strategies in Total Synthesis. I have also supervised undergraduate and graduate students in lab courses, during research projects, and master theses projects.
Source Code and Datasets
- QM-GNNIS is a machine-learned implicit solvent model for quantum chemical calculations that I developed with my colleague Paul Katzberger.
- The second generation of the AMP architecture developed together with Moritz Thürlemann.
- A neural network potential interface for the GROMOS molecular dyanmics engine implemented in C++. The interface was designed to bring AMP and QM-GNNIS among other models from Jupyter notebooks to high-performance calculations.
- To enhance the functionality of GROMOS to perform QM/MM MD simulations, I implemented interfaces to the QM software packages xtb and ORCA. I also updated the accompanying GROMOS++ command line tools.
- We trained the second generation of the AMP architecture on three large and high-quality datasets of QM energies, gradients, and other properties derived from QM/MM electrostatic embedding simulations.