What we do
Developing robust and reliable open-source software tools tackling some of the toughest challenges in cancer research and computational biology is key to our work. For an overview, see below or
Reconstructing cancer phylogenies from copy-number alterations
MEDICC2 is the leading software tool and algorithm for reconstructing cancer evolution from haplotype-specific (or total) somatic copy-number alterations (SCNAs). MEDICC2 is based on a minimum evolution criterion. It measures the distance between any pair of SCNA profiles as the minimum number of whole-genome doubling events, and gains or losses of arbitrary length, needed to transform one genome into another.
- Infers the tree topology that explains the data best according to the above criterion.
- Reconstructs the unsampled ancestral genomes at the internal nodes of the phylogenetic tree.
- Determines and extracts the number of evolutionary events along each branch of the tree.
MEDICC2 is available on Bitbucket at https://bitbucket.org/schwarzlab/medicc2.
Phasing of somatic copy-number alterations
Refphase implements multi-region reference phasing for assigning somatic copy-number alterations (SCNAs) to their parental haplotype-of-origin. Refphase was the algorithm with which Mirrored Subclonal Allelic Imbalance (MSAI) events were discovered in the TRACERx-100 cohort (see here), and which allowed us to identify continuous structural and parallel evolution across human cancers (see here).
- Assigns SCNAs to their parental haplotype-of-origin, revealing parallel evolution and MSAI in multi-region sequencing scenarios.
- Analyses multi-region tumour samples and stratifies individual copy-number segments relative to ploidy into concrete event classes.
- Allows systematic comparison of different sample subgroups (e.g. metastases vs primaries) to identify genetic commonalities and differences.
Refphase is available on Bitbucket at https://bitbucket.org/schwarzlab/refphase.
Simulating cancer evolution of one billion cells with spatial constraints
SMITH is a novel tool and algorithm for simulating cancer evolution including random mutations with varying fitness effects. SMITH is based on the classic branching model of cancer which it extends with "confinement", a new mechanism which simulates spatial constraints by limiting clonal growth based on its size.
- Simulates cell growth and cancer evolution for tumours of realistic sizes with more than one billion cells.
- Implements the confinement mechanism which simulates spatial constraints.
- Recreates the rich clonal dynamics only known from explicitely spatial tumour growth models.
SMITH is available on Bitbucket at https://bitbucket.org/schwarzlab/smith.
Fish plots for Python
PyFish is a Python library for creating beautiful Fish (or Muller) plots in Python, which display the clonal dynamics of a tumour cell population, i.e. the size of subclones as they arise over time during cancer evolution.
- Creates Fish plots from user input data, which can be smoothed or displayed raw.
- Is highly customisable with user-defined rendering parameters and color schemes.
- Allows for high-resolution export of figures in publication quality.
PyFish is available on Bitbucket at https://bitbucket.org/schwarzlab/pyfish.
Incidence-based haplotype estimation and reconstruction from GAM
GAMIBHEAR leverages Genome Architecture Mapping (GAM) data to reconstruct chromosome-spanning haplotypes with high accuracy. GAM is a novel experimental technique based on cryosectioning of cell nuclei followed by sequencing to determine chromatin contacts without ligation.
- Leverages local haplotype fidelity in GAM data to construct a haplotype graph from the co-occurence of SNPs in GAM nuclear slices.
- Formulates the haplotype reconstruction problem as a minimum error correction problem and solves it through efficient heuristics.
- Outperforms Hi-C based haplotype reconstruction in completeness and contiguity, while maintaining similar phasing correctness.
GAMIBHEAR is available on Bitbucket at https://bitbucket.org/schwarzlab/gamibhear.
Visualisation and explorative analysis of multiple-sequence alignments
ALVIS implements Sequence Bundles, a visualisation technique of sequence motifs that maintains horizontal dependencies between adjacent sites in a multiple-sequence alignment (MSA).
- Provides an interactive graphical user interface to explore large MSAs and visualises them using Sequence Bundles.
- Leverages Fisher score embedding of sequences using a Profile Hidden-Markov-Model to enable quantitative analyses.
- Aligns phylogenetic trees to the alignment to allow identification of phylogenetic signals in MSAs.
ALVIS is available on Bitbucket at https://bitbucket.org/schwarzlab/alvis.