Ideas for Thesis Projects

The following is an incomplete list of ideas for thesis topics in (Algorithmic) Bioinformatics that may be supervised by members of the group. You are encouraged to suggest your own ideas and follow your own interests as well.

Polishing circular assemblies from long reads

Most long-read assemblers are optimized to assemble linear, not circular, contigs. Some assemblers do support circular structures, but the resulting contigs may need polishing. The goal of this project is to investigate different strategies for refining contigs produced by different assemblers (like CANU) when we expect circular contigs.

In the long run, this project may be extended to write an optimized assembler for circular structures.

Contact: Sven Rahmann

Improving and porting and scientific software to just-in-time compiled Python

A large fraction of scientific software in the biosciences is written in C/C++ (especially numerical code and simulation-heavy code) or in R (especially high-level statistical methods).

The numba project provides just-in-time compilations capabilities for Python and is working towards a v1.0 release. The advantage is jit compilation is that the compiler can exploit and optimize for what we might call “runtime constants”, i.e., parameter values that are known only at runtime, but that do not change once they are initialized. If we compile the time-critical code after the initialization work has been done, we may achieve speeds beyond what is possible with C++ or other statically compiled languages.

We offer seveal topics that aim to port existing scientific software (or part of its functionality) to the Python universe, extending the functionality, improving the user experience and optimizing the speed all at the same time.

Such a project makes most sense together with another group that relies on compute-intensive software that has proven to be a time-bottleneck in their research workflow.

You will need good Python programming skills, as well as a good unterstanding of low-level (C or assembly) code, e.g., bit manipulations. Essentially you will be writing part high-level Python code using existing scientific libraries (numpy, scipy, pandas, ML packages) and part low-level C code, but using Python syntax.

Contact: Sven Rahmann, Vu Lam Dang, Jens Zentgraf

A modern sequence and alignment editor

Sequence management, alignment editing and figure generation is an important task performed daily by biologists. Older existing tools are still widely used but no longer maintained. We propose the creation of a new open-source software package that aims to replace and extend them.

View Details.

Contact: Vu Lam Dang

Alignment-free analysis of whole genome data

We run several projects that aim to analyze whole genome sequencing (WGS) data using alignment-free methods: We look at short DNA fragments (so-called “k-mers”) and their counts in each sample of a large dataset to answer questions about (for example) copy number aberrations, structural variants, disease-specific genome alterations, etc.

The basis for these methods is an extremely efficient hash table implementation that we have developed recently. In each of the thesis projects offered here, we will explore a different genome-scale application of this technology and develop new algorithms to solve genome analysis questions.

This includes the analysis of real datasets, either publically available ones, or exclusive data provided by partners from the Faculty of Medicine or Biology.

Contact: Sven Rahmann

Fast PWM matching using just-in-time compilation

Fifteen to twenty years ago, the first algorithm appeared that allowed fast matching of position weight matrices (PWMs) against genomes. These algorithms have since been further engineered and the resulting tools have become quite sophisticated, e.g. MOODS. We would like to find out whether the speed of PWM matching can be further improved using just-in-time compilation techniques when the matcher is compiled after the motif is known.

Contact: Sven Rahmann

Algorithmic Bioinformatics, SIC, Saarland University | Privacy notice | Legal notice