Prof. Dr. Achim Tresch

Research Area: Computational Biology, RNA Biology

Branches: Computational BiologyMolecular Biology

Website: Tresch Lab

1. Research Background:

Statistical data analysis and machine learning, single cell and spatial omics technologies, RNA metabolism

2. Research questions addressed by the group:

1. Single-cell and spatial omics technologies

Single-cell RNA-Seq was the first technology that allowed us to monitor gene regulatory processes on the level of the individual cell. Since then, scRNA-Seq has been complemented by other single-cell omics technologies (e.g., scATAC-Seq), and recently single-cell data can be augmented by spatial information. These developments pose several significant bioinformatics challenges: How can we meaningfully combine information from two different single-cell experiments if only one modality can be measured per experiment? How do we reveal biologically relevant, systematic spatial patterns in high-dimensional, noisy data [1, 2, 3]?

2. Models of RNA metabolism

RNA and protein homeostasis are essential components of cellular integrity that deteriorate with age. Using RNA labelling techniques, we could measure RNA synthesis- and degradation rates at a genome-wide scale [4]. We further measured Polymerase II elongation and initiation speed, which are critical steps in RNA synthesis [5]. Currently, we apply our techniques to quantify RNA synthesis, RNA export, and nuclear and cytosolic RNA degradation on a genome-scale. By targeted interventions into the nuclear export system, we want to uncover the mechanisms that lead to RNA export or nuclear retention.

3. Possible projects:

1. Single cell and spatial omics technologies

  • Integration of two single cell omics modalities of the same population of cells
    Many groups use neural networks (typically autoencoders) to map two kinds of single-cell data to a joint latent space. We believe that we can do that better (in particular unbiasedly) with a graph-embedding algorithm, borrowing some concepts from the famous tSNE algorithm.
  • Maturation of olfactory cells.
  • Olfactory receptor (OR) genes are a large group of evolutionarily conserved genes (ca. 350 in humans) whose products act as chemosensors. Apart from their primary role in the olfactory epithelium, ORs are also expressed in various other cells and tissues, such as immune cells. Strikingly, most mature OR-expressing cells display exactly one OR receptor. We want to understand how a cell’s microenvironment influences OR selection, and how specific ORs sense local injuries in the heart. This project will perform spatial transcriptomics in conjunction with single-cell RNA-seq of healthy and inflamed heart valves.
  • Spatial Components Analysis
    Principal Components Analysis (PCA) is one of the oldest and still most powerful methods to project high dimensional data onto a low dimensional space while maintaining pairwise point distances as good as possible. We develop a similar method that projects spatial data onto a few components that conserve the spatial patterns in the data. This will allow us to automatically identify active gene sets in specific tissues or cell clusters.

2. RNA metabolism

  • Mechanistic dissection of the RNA export machinery
    Using our recently established protocol to measure nuclear RNA export, we will target the following questions:
    • What is the contribution of nuclear export to RNA quality control? How much RNA does not pass the filter of nuclear export and how much RNA vanishes through nuclear/cytosolic decay?
    • Which RNA populations are held back in the nucleus, and why? What are the sequence patterns / folding patterns/protein binding patterns that are recognized by the export machinery?
    • How does quality filtering change with age and under environmental changes?

4. Applied Methods and model organisms:

We exploit the whole toolbox of statistical and machine learning methods and adapt them for our purpose, such as dimension reduction techniques, clustering methods, hidden Markov models, supervised techniques (e.g. generalized linear models, support vector machines, neural networks).  This requires fluency in optimization and sampling methods, like (stochastic) gradient descent, Expectation-Maximization, variational inference, Markov Chain Monte Carlo.

5. Desirable skills and qualifications:

  • A background in Statistics / Mathematics / Physics / Computer Science / Computational Biology
  • Experience with R, Python, Matlab etc. A high awareness of quality in statistical data analysis (reproducibility, robustness, interpretability)
  • Interest in interdisciplinary research, interest in modeling biological processes

6. References:

  1. Anna Katharina Schlusche et al. Developmental HCN channelopathy results in decreased neural progenitor proliferation and microcephaly in mice. PNAS 2021. https://doi.org/10.1073/pnas.2009393118
  2. Henrik Failmezger et al. Computational Tumor Infiltration Phenotypes Enable the Spatial and Genomic Analysis of Immune Infiltration in Colorectal Cancer. Front Oncol. 2021 https://doi.org/10.3389/fonc.2021.552331
  3. Mohammad Hussainy et al. Pseudotime analysis reveals novel regulatory factors for multigenic onset and monogenic transition of odorant receptor expression. M Hussainy, SI Korsching, A Tresch. BioRxiv, 2022. https://doi.org/10.1101/2022.01.31.478392
  4. Christian Miller et al. Dynamic transcriptome analysis reveals dynamics of mRNA synthesis and decay in yeast. Molecular Systems Biology, 2011. https://doi.org/10.1038/msb.2010.112
  5. Björn Schwalb et al. TT-seq maps the human transient transcriptome. Science 2016. https://doi.org/10.1126/science.aad9841