Researchers from 14 organizations have proposed a new quality standard and curation process for computational models of biological systems. Writing in a perspective in the December issue of Nature Biotechnology, the group describes MIRIAM – minimum information requested in the annotation of biochemical models – and argues its adoption “will enable users to have confidence that curated models are an accurate reflection” of their associated description.
Computational modeling of biological systems has grown in recent years, driven by the flood of ‘omics experimental data, affordable computational resources, new modeling languages such as the Systems Biology Markup Language (SBML), and emergence of public databases such as BioModels Database and CellML Model Repository.
A handful of companies – Entelos, Gene Network Sciences, Genstruct, and Optimata, are examples – provide modeling technology for drug discovery and development while a growing contingent of academic researchers are developing and working with models. Model sophistication varies dramatically: Some are static descriptions while others permit wide-ranging bio-simulation.
Click here for a free
Bio-IT World briefing
on Systems Biology
The diversity of modeling approaches, languages, and documentation has hampered the sharing of models and complicated efforts by peer reviewers evaluating research involving the use of models. Speaking at a recent roundtable of modeling held by Bio-IT World, one of the perspective’s authors, researcher Herbert Sauro of the Keck Graduate Institute, noted it often takes two weeks to curate models submitted to BioModels.net because of errors made during the submission process.
MIRIAM advocates hope their proposed standard can do for modeling what MIAME (minimum information about a microarray experiment) is doing for microarrays, which is build confidence and advance cross-comparisons. The need is especially great for quantitative models.
“Databases of quantitative models are valuable resources only if researchers can trust the quality of their content. Similarly, repositories are not useful unless users can search for specific models and then relate model constituents to other data sets such as bioinformatics databases and controlled vocabularies,” wrote the authors in the Nature Biotechnology perspective.
MIRIAM has two parts: 1) a set of checks that match a model to its description (‘reference correspondence’, often a publication in a scientific journal), and 2) a set of ‘annotation schemes’.
The first of these documents the model’s provenance, i.e. who created it, whether it’s been modified, and a stable link to its full description.
The second scheme links the components of the model to relevant bioinformatics resources. For example, a model of alcohol metabolism in the liver would be annotated with links to the protein databases for all the enzymes involved in this pathway, and database links to all the relevant metabolites.
The goal, they say, is to make it easier for researchers to search models on the basis of their components, to contact the creators of the model if they need more information, and to track the history of a model if it has been modified. MIRIAM’s creators include representatives of four major repositories for models (BioModels Database, CellML Model Repository, DOQCS, and SigPath), all of which are now in the process of making the models in their repositories MIRIAM compliant.
“By adopting MIRIAM as a voluntary code of conduct, we will be able to provide our users with a reasonable level of quality assurance, so they’ll be able to get on with the business of generating and testing new hypotheses instead of recoding someone else’s old hypothesis,” says one of the perspective’s authors, Nicolas Le Novère, European Bioinformatics Institute.
“We also hope that journal editors will adopt MIRIAM as a quality control measure for papers that describe models. This approach has worked very well for other fields – for example the microarray community, by enabling authors, publishers, and data providers to work together to improve access to meaningful biological information,” says Le Novère.
How relevant MIRIAM will become to commercial efforts is not yet clear. A good deal of IP, such as experimentally verified kinetics, is usually contained in their models. Also, MIRIAM-compliance is intended to verify whether a model is well documented; it won’t demonstrate the model accurately reflects nature. If journals heed the call, the new standard would be a step forward.