This page will be updated starting in the Spring of 2020, as the EDBB program will be informed of new positions becoming available for the June 2020 Hiring Days event at EPFL. Meanwhile, do not hesitate to contact the laboratories which interest you to find out whether they have upcoming openings for PhD students.
Due to the Coronavirus, the June 2020 Hiring Days will take place remotely. More information to follow here.
Vassily HATZIMANIKATIS – Laboratory of Computational Systems Biotechnology / Institute of Chemical Sciences and Engineering / School of Basic Sciences
Understanding the cellular processes is crucial for making progress in medicine, biology, and biotechnology. These fields aim to characterize the behavior of the cell under different conditions to provide tools for personalized and precision medicine, green energy or efficient chemical production. Experimental approaches are currently generating an abundant amount of biological data and require computational methods to perform integrative analysis of the cellular processes.
In the Laboratory of Computational Systems Biotechnology, LCSB, we focus on cellular process modeling, large-scale computations, and data analysis, with the aim to develop mathematical models and novel methods of mathematical and computational analysis for e.g. systems biology, metabolic engineering, and prediction of novel biotransformations.
We have openings for two to three PhD positions with an expected starting time-frame of Fall 2020. The following research topics are offered:
- Machine learning techniques for designing large-scale and genome-scale kinetic models
This project aims to employ Generative Adversarial Neural networks (GANs), an emerging technique in the area of deep learning, to construct populations of large-scale and genome-scale kinetic models of metabolic processes in cellular organisms. The emphasis of model design will be on obtaining models that are consistent with the experimental observation and consistent with imposed criteria such as biological feasibility, cell bioenergetics, and other physicochemical constraints.
2. Development of dynamic and hybrid models of cellular metabolism
The aim of this project is to develop kinetic models of of a model eukaryotic organism, S. cerevisiae, that will allow us to predict the metabolic responses to changes in cellular and process parameters. In the next phase, the candidate will improve further the predictive capabilities of models by integrating the protein expression system and regulatory interactions of this organism. This project is part of an EU project and will involve interdisciplinary collaboration with the experts in systems biology, synthetic biology, and metabolic engineering.
- Microbiome data analysis and modeling
In this project, we aim to develop mathematical models that describe the metabolic networks of individual organisms in microbial communities and the interactions through metabolites and competition for resources. We will also develop individual agent-based representations of bacterial motility and growth using adaptive metabolic networks for each agent-cell and study how metabolic interactions can give rise to spatio-temporal arrangements in microbial communities.
At least 2 openings are available for the April 15th, 2020 deadline. PhD projects will be available in the following two areas of research:
- Biological nanopores for single-molecule sensing
Nanopore sensing is a powerful single-molecule approach currently developed for the precise detection of biomolecules, as for instance in DNA and protein sequencing. Our laboratory is developing this technology exploiting the properties of biological pores. Recently, we showed that aerolysin, a pore-forming toxin, exhibits high sensitivity for single-molecule detection and can be ad hoc engineered for different sensing tasks. The goal of this project is to develop and characterize aerolysin-based nanopores as sensing devices to be applied for genome sequencing, proteomic analysis and disease diagnosis. The project is highly interdisciplinary, includes experimental and computational aspects and interactions with a diverse network of collaborators. Students with a background in biochemistry, physics, bioengineering and computational sciences are encouraged to apply.
- Integrative modeling at the membrane-protein interface
Molecular interfaces are essential for the formation and regulation of all assemblies that sustain life, to define cellular boundaries and intracellular organization, and to mediate communication with the outer environment. Our laboratory has been studying the molecular mechanisms governing the association of proteins to their membrane interfaces in order to understand the functional implications of this interplay. Multiple projects are available that focus on the theoretical and computational investigation of the structural and dynamic properties of membrane protein systems. All of them are addressed in synergy with experimental collaborators to allow for an efficient integration of biochemical and biophysical data. Students with a background in biochemistry, physics, bioengineering and computational sciences are encouraged to apply.
We have at least two possible projects for PhD students:
1) Biological systems are often said to be ‘robust’. This means that when they are perturbed, for example, by changing environmental conditions, they do not just stop working and break down. However, robustness certainly has limits, and we would like to explore those, because when systems are no longer robust, they are easier to manipulate. Robustness may be good biologically but it can be bad for us wanting to modify the system. Also, in some diseases, genes that were needed for robustness may be corrupted and we may be able to specifically target such cells using their non-robustness, which we call fragility. This project involves experiments involving yeast molecular biology, microcopy, convolutional neural networks, as well as theory to develop these ideas.
2) A goal of neurosciene is to predict an animal’s behavior by understanding its brain. However, for almost no animal can we actually do that well. Suppose we take one of the simplest brains studied in the lab (belonging to the roundworm C. elegans), we stimulate it as we want experimentally, and at the same time look at most of its neurons, as well as measure its behavior. Can you come up with theory and experiments to predict what the worm will do a few seconds later? This research involves C. elegans molecular biology, microscopy, microfluidics, convolutional neural networks, and theory.
The Suter lab is interested in quantitative analysis of gene expression to understand how cell identities are established and maintained. The PhD project we propose aims at quantitative, biophysical characterization of the transcription factor network that controls the identity of embryonic stem cells. It will involve cutting edge approaches such as genome editing, quantitative live cell imaging and cell tracking, and single molecule imaging. This project is part of a Sinergia Consortium and will involve interdisciplinary collaboration with our partner labs experts in microfluidics and in vitro transcription factor characterization (Maerkl lab, EPFL), and computational modelling of biological networks (van Nimwegen lab, University of Basel).
We also propose a project to study the role of mitotic bookmarking in cancer stem cell self-renewal. Cancer stem cells are central to the fueling of tumorigenesis through their ability to self-renew. Over the past years, transcription factors binding to mitotic chromosomes have been suggested to play a role in the ability of stem cells to self-renew, but whether mitotic bookmarking plays a role in self-renewal of cancer stem cells is unknown. Here the candidate will explore the role of mitotic retention of oncogenic transcription factors in the ability of cancer stems cells to maintain their gene expression program over cell division. To tackle this question, the PhD candidate will learn and apply a broad set of approaches, such as live cell fluorescence microscopy, genomics approaches (ChIP-seq, CUT&RUN, ATAC-seq, RNA-seq), genome editing using CRISPR technology, optogenetics, and in vitro 3D culture and migration assays.
1) Characterization of spatiotemporal organization of the brain lipidome
Neural cells produce thousands of different lipids, each endowed with peculiar structural features and contributing to specific biological functions. Lipid composition affects neuron firing properties influencing vesicle fusion and fission processes, membrane conductivity, and ion fluxes. Nonetheless, a systematic and fine-grained characterization of lipid composition in the different brain regions is not available.
Lipids also play a fundamental role in brain development. For example, some lipids, such as glycosphingolipids, mediate cell-cell recognition, others like steroid hormones, and phosphoinositides, have a role in stimulating cell growth and signaling. Furthermore, exposure to teratogenic agents, during development, is associated to cognitive or sensory impairments that might be mediated by interference of these teratogens with lipid biogenesis and metabolism. However, little is known about how the regional specificity of lipids is developmentally established and maintained throughout adulthood.
The doctoral candidate will aim at filling this gap by collecting systematic data necessary to construct a high spatially resolved atlas of the lipidome of the adult and developing mouse brain. We expect this resource to provide numerous cues of the underlying regulation mechanisms; the most interesting observations will be experimentally followed up by the candidate and related to function.
The project offered jointly by the La Manno and D’Angelo labs will allow the candidate to:
- Use super-resolved Imaging Mass Spectrometry (IMS) to reconstruct the spatial lipidome in serial brain sections from adult and developing
- Assess the lipid deregulation resulting from the exposure of different teratogenic conditions.
- Investigate the relation between the lipidome of different stem cell populations and their neural
- Investigate how perturbation to genes involved in lipid metabolism affects brain development.
- Assess how direct perturbations of lipid composition affect morphogenesis and adult brain structure and composition.
2) Construction and analysis of a Lipid Brain Atlas
Single-cell and spatial transcriptomics technologies have matured significantly in the last few years and are now extensively used to build comprehensive atlases of tissue gene expression heterogeneity. However, while datasets of this kind are accumulating, similar resources that describe biochemical heterogeneity of tissues are still lacking.
With the advent of super-resolved Imaging Mass Spectrometry, it is now possible to efficiently and rapidly measure the biochemical composition of tissues at micron- resolution. Using the technique, the laboratories of Giovanni D’Angelo and Gioele La Manno have recently found a substantial spatial organization of lipids in the brain, the regional specificity found was significantly more extensive than previously believed.
In the brain, the role of lipids is crucial for different functions; for example, it contributes to setting neuron firing properties, controls membrane conductivity and ion fluxes. Analyzing this unexplored heterogeneity is likely to reveal new biochemical processes and principles that characterize different neurons and brain areas. Our labs are actively working to collect an extensive dataset to build a resource that will serve as a powerful tool for neurochemical research.
The doctoral candidate will have a central role in this effort. She/he will develop new computational methods to process, analyze, and organize this extensive dataset. We expect the candidate to:
- Develop ad-hoc machine learning algorithms (e.g., latent variable decomposition, deconvolution approaches) for the analysis and interpretation of spatial lipidomic
- Use the tools developed to analyze the regional heterogeneity of the mouse brain lipidome.
- Organize this knowledge into a resource for the
- Construct a volumetric 3d model of the brain and register all the data obtained in this reference
- Integrate the lipid atlas with gene expression brain atlases and build principled models that can predict one data type from the other.
In the Neuroengineering Laboratory, we investigate transgenic flies (Drosophila melanogaster) to understand how behavior is controlled and to design more intelligent robots.
We have several PhD openings to study one of the following interdisciplinary areas:
(1) Data-driven neural and biomechanical modeling of limb control.
Key techniques: Computational modeling, Simulations, Confocal microscopy, Genetics
(2) Optical recordings of neuronal population dynamics for limb control.
Key techniques: 2-photon microscopy, Machine learning, Genetics
Join us! There is much to discover!”
- Investigation of centriole number control mechanisms in human cells
Centrioles are small organelles that are critical for forming cilia, and which exhibit a striking 9-fold radial symmetric arrangement of microtubules. In proliferating cells, centrioles assemble once per cell cycle next to an existing centriole, reaching a length of ~500 nm. Three proteins are critical for the onset of centriole assembly in human cells: the kinase Plk4, as well as the coiled-coil containing proteins HsSAS-6 and STIL. Depletion of either component prevents centriole assembly, whereas their overexpression results in excess centrioles.
Combine experiments –depleting proteins or providing them in excess- and mathematical modeling to investigate how Plk4, HsSAS-6 and STIL ensure that a single centriole assembles next to each existing centriole. Investigate also mechanisms by which these components ensure that centrioles achieve a characteristic length.
Molecular biology, including CRISPR/Cas9 genome engineering, cell biology, live cell imaging, Fluorescence Correlation Spectroscopy (FCS), super-resolution microscopy (STORM, iSIM, Ux-EM-STED), mathematical modeling.
- Exploring evolutionary diversity and origin of SAS-6 proteins
Centrioles are small organelles that are critical for forming cilia, and which exhibit a striking 9-fold radial symmetric arrangement of microtubules. We discovered that the evolutionarily conserved SAS-6 proteins self-assemble into 9-fold radially symmetric structures thought to template the formation of the entire organelle.
Develop machine learning strategies (e.g. using Tensorflow) for high throughput protein structure prediction. Apply machine learning to identify homologues of SAS-6 across all domains of life and thus help trace its origin. Proceed likewise for other proteins critical for centriole assembly and thus eventually trace organelle origin.
Computational biology, machine learning, structural prediction; collaboration with the Dessimoz laboratory (UNIL, SIB).
Group website: https://www.epfl.ch/labs/bitbol-lab/
(1) Understanding how optimization and phylogeny shape protein sequences
Proteins play crucial roles in our cells. The amino-acid sequence of a protein encodes its function, including its structure and its possible interactions. In evolution, random mutations affect the sequence, while natural selection acts at the level of function. Shedding light on the sequence-function mapping of proteins is central to a systems-level understanding of cells, and has far-reaching applications in synthetic biology and drug targeting. The current explosion of available sequence data enables data-driven approaches to discover the principles of protein operation.
In alignments of homologous protein sequences, correlations exist between certain amino-acid sites. We aim to establish a full decomposition of these correlations, dissecting signatures from functional optimization, and from evolutionary history. We also aim to make new predictions for protein-protein interactions from sequence data, and to understand whether real proteins are mechanically optimized.
Several directions are possible, depending on the background and tastes of the student:
– Theoretical directions involve developing statistical physics based simulations to generate controlled synthetic data, and employing analytical calculations to make sense of the synthetic data.
-Data-driven directions involve analyzing real protein sequence data, as well as protein structure data.
(2) Characterizing the exploration of rugged fitness landscapes by subdivided populations
Populations of living organisms are pushed toward optimality by natural selection. However, from a given state, populations may not be able to reach the absolute maximum of a fitness landscape, which represents fitness (reproduction rate) versus genotype (genetic type). Indeed, fitness landscapes are often rugged, like the energy landscapes of glassy physical systems. Real microbial populations are subdivided between habitats, e.g. among different organs and among different hosts in the case of an infection. This may help microbial populations to better explore fitness landscapes.
We aim to characterize how subdivided microbial populations explore rugged fitness landscapes, and to establish a universal description of subdivided populations on graphs. We also aim to study the impact of population subdivision on the evolution of antimicrobial resistance, and to investigate how expanding populations explore rugged fitness landscapes.
Several directions are possible. They all involve a combination of analytical calculations and numerical simulations, with proportions depending on the background and tastes of the applicant:
– Studying biased random walks on rugged fitness landscapes and models of populations on graphs.
– Developing applications to antimicrobial resistance evolution and expanding populations.
Jaksic lab is recruiting a PhD student to work on genetic basis and evolution of cognitive ability and related neuronal traits.
Our lab is a new lab with new ideas. Our mission is to merge the fields of experimental evolutionary biology and neuroscience by developing high-throughput integration of technologies from both fields with the end-result of experimentally evolving a cognitive brain in a model organism. We are specifically interested in exploring and mapping genetic variation underlying complex behavioral traits such as cognition in Drosophila using experimental evolution, next generation sequencing, high-throughput imaging techniques and complex behavior phenotyping. These technologies highly rely on successful integration of computational and quantitative approaches such as bioinformatics, machine learning (and other statistical approaches), automatization, and real-time image data analysis with experimental methods such as high-throughput phenotyping, robotics and efficient and creative experiment designs.
We are looking for highly motivated students with a good background in computational and quantitative skills (programming/scripting experience in languages such as Python, Matlab, C++, Java, R or similar) and with strong interest in animal behavior, evolution, genetics, or neuroscience.
The project you will be working on will heavily rely on your computational skills but is, in essence, highly multidisciplinary.
It will be based on design and automatization of high-throughput phenotyping of
– Various complex behavioral traits in a diverse genetic panel of Drosophila using automated real-time video tracking with implementation of machine-learning-based decision making and selection algorithms,
– Neuronal morphology in Drosophila using high-throughput imaging and image data analysis of fluorescently labeled neurons and other brain tissues,
and quantitative and computational analyses such as
– Genotype-phenotype mapping using whole-genome sequencing data,
– Generation and analysis of time-series, whole-genome sequencing and transcriptomics data,
The project, especially the experimental part, will be highly collaborative, and you will have assistance and guidance of other lab members. Additionally, through the design and development of automatized phenotyping algorithms you will have an opportunity to participate in the set-up of the first long-term evolution experiment for selection on cognitive ability. You will have a chance to generate and analyze time-resolved whole genome sequencing data that will enable us to observe and track real-time evolution of the brain from DNA to phenotype level for the first time ever.
Your position will be 50% experimental work (experimental setup for streamlined behavior data collection using computational approaches such as real-time behavior classification using machine learning) and 50% computational (bioinformatics, data analysis), however both will require creative and quantitative thinking, computational skills and interest in biology of behavior.
You can expect to develop and improve your bioinformatic and computational skills, but also learn population genetics, and quantitative techniques in evolutionary biology, gain knowledge of Drosophila genetics and neurobiology, and become an expert in experimental evolutionary neurobiology.
You can expect a supportive, inclusive, collaborative, dynamic and fun research environment, open-door mentorship, flat lab hierarchy, opportunity to attend international conferences, and access to the academic network of evolutionary biology.
You can learn more about the lab, projects and your future PI at jaksiclab.com.
If you think you would like to join our team and become a pioneer in experimental evolutionary neurobiology, do not hesitate to contact me!
Ana Marija Jaksic
The Weigert lab is a computational group that focuses on data-driven image analysis for biological applications. In particular we are interested in the development of new machine learning based approaches to extract quantitative biological information from large microscopy data sets, and novel computational methods to augment and improve optical microscopy.
There are several research projects available, covering the topics of
i) ML-based segmentation of large scale 2D/3D microscopy images
ii) Data efficiency and self-supervised learning for microscopy
iii) Reconstruction methods for novel microscopy modalities
If you are fascinated by the richness of biological images, are interested in mathematical/physical problems, enjoy programming (e.g. Python, C++, Cuda, …), and share a general interest in the design and practical implementation of machine learning based vision methods, please get in touch!