Studying human phenotypes with mice

7th February 2019

The International Mouse Phenotyping Consortium (IMPC) measures associations between mouse genes and phenotypes. Many of these genes have orthologs in other animals and the phenotypes have analogs across species too. These similarities make it possible to learn from observations in mice about the causes of human genetic disease.

A key step in translating findings in mice to humans is mapping phenotypic annotations across the two species. This calculation begins from disease features codified using the Human Phenotype Ontology (HPO). The human phenotypes are then matched to mouse phenotypes encoded using the Mammalian Phenotype Ontology (MPO), making it possible to identify relevant IMPC measurements.

The widget below summarizes the results of this calculation, showing the closest mouse phenotype for a particular human phenotype, as well as the most relevant term measured by an IMPC protocol. As an example, consider ‘Ataxia‘, which is a neurological condition characterized by difficulties with movement. Enter this term in the search box and select it from the drop-down list. A table underneath will reveal that the best match in the mammalian ontology is ‘ataxia’ – an equivalent term. The most relevant IMPC protocol is for ‘abnormal gait’.

What human phenotypes are you interested in? Type them into this interactive widget to see how they might manifest in the mouse and how they might be captured in the IMPC pipeline. Examples: ‘Ataxia’, ‘Glucose intolerance’, ‘Vocal cord dysfunction’.

The widget displays numerical scores for each phenotype match. These provide hints about how informative each match is based on proximity in the ontology structure and background phenotype frequencies. The absolute scores are not immediately intuitive, but their relative values are instructive. The higher the score, the closer the match and rarer the phenotype and, thus, the more powerful it can be for establishing a genotype-phenotype relationship.

In our ataxia example, the score linking to mouse ‘abnormal gait’ is lower than the score matching with mouse ‘ataxia’. This indicates that the IMPC protocol captures some, but not all, aspects of the disease phenotype. This is very common – a phenotypic screen cannot in practice measure all possible aspects of all diseases. Nonetheless, the partial match indicates that an IMPC measurement can provide supportive evidence. In other cases, IMPC measurements provide direct assays for a phenotype, e.g. ‘glucose intolerance’. In yet other cases, a human phenotype may not match with any mouse phenotype, e.g. ‘vocal cord dysfunction’, or match only with very low scores, e.g. ‘cellulitis’, revealing a limited ability to model that feature.

Diseases are often characterized by several features that occur together. You can enter more phenotypes into the widget to see scores for your phenotypes of interest. However, the simple widget does not perform any integration. For that, the IMPC web portal provides holistic comparisons with diseases from OMIM, ORPHANET, and DECIPHER. Results are provided for over 5,000 IMPC mouse models and 18,000 disease definitions, see e.g. all diseases involving ataxia.

Widget constructed using React. Icons from Font Awesome (modified).

Leave a Reply