From Machine Learning to Mechanisms: Functional Predictions in Genomics and Medicine

Scalable models for Molecular Biology

MolecularBioModelZoo (MBMZ): scalable and reusable models for molecular biology

Partners Involved: Gagneur*, Stegle

The aim is to develop a framework program for models in computer-based biology that significantly simplifies both the reuse and integration of different models. Special emphasis will be put on:

i) Comparability: models of the scientific community can be stored and directly compared in a uniform platform using the MolecularBioModelZoo (MBMZ);

ii) scalable applicability of models to new data sets: users can use stored models, for example, to interpret genomic variants in patients, whereby automatic mechanisms for parallelization of programs on mainframes help to deal with very large data sets;

iii) modular connection of existing models to new models: existing models can be connected to larger and more complex units.