Department Bionanoscience
Principal investigator Nikolina Sostaric
E-mail address n.sostaric@tudelft.nl
Website https://nikolinasostaric.github.io/
Aptamers as biosensors
Suitable as a BEP? Yes
Suitable as a MEP? Yes
Suitable as an Academic Research Project? Yes
Techniques:
- Programming
- MD simulations
- 3D biomolecules modelling
Single-stranded DNA aptamers serve as bioreceptors in wearable biosensors and allow for continuous health monitoring of humans. In this computational project, we will investigate the protein:aptamer interactions to lay the foundation for exciting biosensing applications. To better understand the details of this interactions, we will simulate the aptamer and protein:aptamer complex with molecular dynamics simulations, and try to identify crucial parameters that influence this system, such as temperature, salinity of the buffer, and more.
During this project, you will gain experience with modelling of biomolecules, molecular dynamics simulations of nucleic acids and proteins, and usage of Delft supercomputer. You will also develop a better understanding of biomolecular structure and function, and develop data analysis skills.
The project will be co-supervised by Lena Fasching.
Further reading
A publication that we’re currently preparing on this topic can be shared with interested students
Rhodopsin proteins as voltage sensors
Suitable as a BEP? Yes
Suitable as a MEP? Yes
Suitable as an Academic Research Project? Yes
Techniques:
- Programming
- MD simulations
- 3D biomolecules modelling
Rhodopsins are fluorescent membrane proteins that absorb photons and use their energy to power proton transport. 15 years ago, it was discovered that the fluorescence intensity of rhodopsin proteins is dependent on membrane voltage. This allows us to use them as tools to detect electrical signals in neurons via light microscopy, which has revolutionized the field of neuroscience. A lot of research has gone into protein engineering of these voltage sensitive rhodopsins. However, the full potential of these proteins for voltage measurement has not yet been reached, partly because the current understanding of how they work is limited.
We have collected a large dataset of rhodopsin mutants of which we know fluorescent properties as well as proton pumping behaviour, and we would like to use this data using computational biology techniques in the fields of molecular dynamics, machine learning and bioinformatics. In this project you will be applying a variety of computational techniques to help understand and engineer voltage sensitive fluorescent proteins. You will learn how to handle large amounts of data and carry out high performance computations on the supercomputer, DelftBlue.
Further reading
A previous MEP thesis can be shared with interested students
Protein interactions in 3D
Suitable as a BEP? Yes
Suitable as a MEP? Yes
Suitable as an Academic Research Project? Yes
Techniques:
- Programming
- MD simulations
- Working with omics data and 3D biomolecules
Protein-protein interactions (PPIs) are essential for functioning of any living cell. While some proteins form strong interactions and bind into permanent complexes, other PPIs are of a transient nature, forming interactions of varying strength and time duration. Moreover, a cell may modulate binding affinity of a given protein:protein pair by, for instance, changing the amounts of proteins or introducing/removing post-translational modifications.
In this project, we will use bioinformatics to explore PPIs in yeast polarity protein network. Our starting point will be the proximity labelling data (TurboID), where interactors of selected polarity network proteins were identified. To distinguish between different types of interactions and to learn more about this network, we will analyze this data using various bioinformatics approaches, including mass spectrometry data analysis, performing simultaneous folding and docking of proteins (hits from TurboID) using AlphaFold2, and prediction of binding affinities, among others.
Further reading
Previous MEP and BEP theses on this work can be shared with interested students
(Example) projects submitted by lab in past years
(2024-2025) The dark interactome
Supervisor: Stefan Loonen, s.loonen-2@tudelft.nl
Protein electron carriers (PEC) are ubiquitous proteins that mediate the electron transfer between small-molecule electron carrier pools and redox-dependent enzymes. Generally an organism contains few PECs (f.e. ferredoxin or flavodoxin) which serve many different enzymes. However, which enzyme gets served by which PEC is not well characterized. We want to investigate what the potential PEC-enzyme pairs are in E. coli, through the use of various computational techniques. The first step will be identifying potential pairs through structure prediction techniques and protein-protein interaction tools such as Alphafold and Foldseek. Depending on the performance of these methods we can further characterize potential PEC-enzyme pairs through machine learning or molecular dynamics based approaches.
Techniques
- Bioinformatics
- Coding
- Protein structure predictions
- Molecular dynamics simulations
- Using DelftBlue supercomputing cluster
Further reading
Xian and Wang 2024
(2024-2025) Simulating properties of elastin-like polypeptides
Supervisor: Stefan Loonen, s.loonen-2@tudelft.nl & Christine Visser, c.m.visser-1@tudelft.nl
Biomaterials are engineered to interact with biological systems, offering significant potential in medicine and nutrition. However, many rely on animal-derived proteins, which limit tunability and raise ethical concerns. This project focuses on sustainably producing natural and engineered proteins like collagen and elastin—key components of the extracellular matrix (ECM)—to develop biomaterials for soft tissue repair and cell-based meat. Collagen provides strength and flexibility, while elastin adds elasticity. To replicate elastin, scientists design elastin-like polypeptides (ELPs), which transition from soluble to insoluble at specific temperatures, crucial for biomedical use. Using molecular dynamics (MD) simulations, we’ll model ELP designs by altering amino acids, linkers, and conditions, identifying promising candidates for further experimental studies.
Techniques
- Molecular dynamics simulations
- Coding
- Usage of DelftBlue supercomputing cluster
Further reading
Barreiro et al. Biomacromolecules 2023, 24, 489−501
(2024-2025) Dynamics of photoreactive proteins
Supervisor: Nikolina Sostaric, n.sostaric@tudelft.nl & Beatriz Orozco Monroy, b.e.orozcomonroy@tudelft.nl
DNA segregation is one of the requirements in building a synthetic cell. To experimentally construct a DNA segregation system for synthetic cells, we are utilizing bacterial proteins known for plasmid segregation fused with photoreactive proteins to enable light-controlled segregation processes
In this project, we would like to acquire a better understanding of the dynamics of these photoreactive proteins and their synthetic constructs, as well as their interactions with DNA. More specifically, we would do 3D modelling and perform molecular dynamics simulations of different LOV2 (light, oxygen, or voltage) domain constructs.
Techniques
- Molecular dynamics simulations
- Protein structure prediction
- Coding
- Usage of supercomputing cluster
Further reading
Peter, E., Dick, B. & Baeurle, S. Mechanism of signal transduction of the LOV2-Jα photosensor from Avena sativa. Nat Commun 1, 122 (2010). https://doi.org/10.1038/ncomms1121
(2024-2025) Clustering data based on activation profiles
Supervisor: Nikolina Sostaric, n.sostaric@tudelft.nl & Marianne Bauer, m.s.bauer@tudelft.nl
We have previously used the information bottleneck algorithm to cluster stationary data and identify groups of protein inputs that fulfill the same function. The goal here is to develop a clustering algorithm based on the number of RNAs in different conditions; this algorithm would need to cluster RNAs that behave similarly across conditions. We would like to see if one can develop an algorithm based on the conditional entropy in this data. This project will rely heavily on analytics and numerics.
Techniques
- Python coding
- Clustering
- Analytics
- Statistics