Finding what's important in our genome

New integrated tool to predict the function of non-coding variants

doi:10.1038/nmeth.2832

Plots showing the performance of the classifier on three different sets of genetic variants used in the study.

Researchers at the Wellcome Trust Sanger Institute and the EMBL-European Bioinformatics Institute have developed software that predicts the likelihood of variants in non-coding regions – relatively unknown regions of DNA that make up 98 per cent of genome – having a functional role.

The software, called GWAVA, integrates an enormous amount of information about the way genes are regulated, and prioritises non-coding variants in the human genome. This helps researchers focus their research on the most promising candidates, potentially saving considerable time and resources.

In recent years scientists have found a lot of links between our genes and susceptibility to disease – but there is still a long way to go before we fully understand how DNA variation underlies disease. While much is known about the way protein-coding genes work, our three-billion-base-pair-long genome is bursting with other types of information. One of the big challenges in genomics is figuring out how non-coding regions of the human genome are involved in disease.

“The information provided by the ENCODE consortium, the 1000 Genomes Project and the NIH’s Roadmap Epigenomics project are extremely useful resources for understanding non-coding variants. But ranking that information is no small task. There is a lot of benign variation in our genome, so we needed a way to narrow down which regions play a role in disease.”

Paul Flicek Co-lead author from EMBL-EBI

The team investigated if a combination of information related to genes, genetic regions associated with regulation and genome-wide properties can be used to identify the most likely variants that contribute to disease in the non-coding part of the genome.

“GWAVA uses a classifier to discriminate apparently harmless non-coding variants from those that are likely to be involved in disease. We tested it out using several scenarios and found that it consistently prioritises the regions known to be associated with disease. This could be really useful for people who need to decide which mutations to look at as cancer drivers, for example.”

Graham Ritchie First author from EMBL-EBI and the Sanger Institute

The authors hope that using GWAVA predictions for non-coding variants in disease association studies will substantially improve the chances of finding genetic variants that are involved in human disease.

“We’ve combined freely available data to predict the impact of these variants in the non-coding region of the genome. Most disease-associated variants discovered to date fall outside genes. This tool can help us start to understand how they work.”

Professor Eleftheria Zeggini Co-lead author from the Sanger Institute

More information

Funding

This work was funded by the Wellcome Trust and by the European Molecular Biology Laboratory.

Participating Centres

• Wellcome Trust Sanger Institute, Hinxton, Cambridge, UK.
• European Molecular Biology Laboratory, European Bioinformatics Institute, Hinxton, Cambridge, UK.