identifying sampling bias in bacterial genome databases

When analyzing the diversity of bacterial populations it is important to consider a potential sampling bias that can distort your pangenome analysis.
Our tool PhyloThin is a coalescent-theory-based approach to identify prokaryotic genomes that can be considered as oversampled.

We are currently writing the manuscript for this tool. If you need preliminary access to PhyloThin, check out our github repository.

ancestral reconstruction of phage infection histories

The immunity memory of the prokaryotic defense system CRISPR-Cas is remarkable, as it encodes a chronological record of past infection attempts in an inheritable spacer array. SpacerPlacer is a tool designed to reconstruct the ancestral states and events that formed this spacer array. If you want to reconstruct the spacer arrays of your favorite species, check out our GitHub repository. We reconstructed the deletion events of spacers in the CRISPRCasdb database, revealing some interesting properties of immunity loss in CRISPR-Cas systems. Have a look at our paper to find out more.

phylogeny-aware detection of gene associations and phage-pangenome interactions

Goldfinder identifies co-occurring and mutually exclusive genes in bacterial pangenomes while accounting for shared evolutionary history. It can also be used to identify interactions between prokaryotic genes and phages.

effective large scale coalescent simulations

msprime uses succinct tree sequences to speed up the simulation of ancestral relationships. We are part of the fantastic scientific open-source community that develops novel features into msprime maintained by Jerome Kelleher. We have implemented the gene conversion mechanism into msprime and currently incorporate additional features that are particularly relevant for microbial evolution.

Have a look at the msprime documentation here.

hierarchical soft clustering for ancestry inference

tangleGen is a hierarchical soft clustering method that leverages ideas from graph theory and machine learning to infer ancestry in complex and admixed populations.

Have a look at what we have done at  github

supervised feature selection for ancestry informative markers

With Peter Pfaffelhuber, Franziska Grundner-Culemann and Veronika Lipphardt we created AIMsetfinder.
AIMsetfinder is a supervised feature selection approach to identify sets of Ancestry Informative Markers (AIMs), that minimize the logloss error of a naive Bayes classifier.

Have a look at what we have done at  github

Pan-genome Analysis and Exploration

Richard Neher, Wei Ding, and I created a pipeline to automatically analyze pan-genomes. The most outstanding part of this project is the phylogeny-based identification of paralogs and the visualization of the pan-genome in a browser. It is now very easy to explore the pangenome and search for certain genes or features within a pan-genome.

Have a look at what we have done at  this demo webpage.