Welcome to the Phenomics and Machine Learning Team in Berlin
Our group has developed the Human Phenotype Ontology and applications of this ontology to differential diagnostics and genomics analyses.
We develop algorithms for phenotype driven analysis of health care information and genomic variation data. We also work in the fields of machine learning and modeling of biological networks with the tools of linear algebra and graph theory.
We create tools for phenotype analyses that make use of the entire clinical phenotyping spectrum, not only using HPO, but also model organisms data (we also create the uberpheno) and other ontologies. We work on tools for acquiring phenotype data from unstructured data sources (EHRs, publications). We have and develop machine learning algorithms helping in medical decision making (e.g. disorder prediction or patient matchmaking), e.g. we have created the Phenomizer. We are working on novel machine learning approaches for phenotype-genotype correlation prediction.
We are partners in software development for the interpretation of genomic variation data such as Exomiser.
The HPO is being developed with the Robinson Group and the Monarch Initiative.