In silico prediction of the disease-causing potential of extragenic DNA variants

• Applicant: Professor Dr. Dominik Seelow
• Subject Area: Human Genetics
• Term: from 2014 to 2018
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 256479157

The rapid development of high-throughput sequencing technologies has revolutionised the elucidation of monogenic diseases. However, the vast number of DNA variants found in each sequencing project impedes the identification of the actual disease mutation. Especially Whole Genome Sequencing approaches put a high burden on the in silico evaluation as millions of variants have to be assessed. Besides, the computer programs most heavily used, PolyPhen-2 and SIFT, can only handle substitutions of single amino acids. Although our software MutationTaster is able to additionally predict the consequences of synonymous and non-coding alterations, it is restricted to intragenic variants and cannot evaluate sequence changes in regulatory elements - which are often located upstream of the regulated gene. So far, none of the common prediction tools is capable to assess extragenic variants. With the proposed project, we aim for the enhancement of MutationTaster's capabilities to allow the analysis of extragenic alterations. We will extend MutationTaster to lift the restriction to intragenic positions and to recognise the interactions of regulatory elements and the regulated genes, based on the bioinformatic evaluation of experimental data. The advance of the ENCODE project has generated a wealth of data for regulatory elements, such as transcription factor binding sites or methylation sites. We intend to use this data along with common polymorphisms from the 1000 Genomes Project and known disease mutations from the Human Gene Mutation Database (HGMD) as training and test cases to develop tests suitable to discriminate harmless from harmful variants in extragenic regions. MutationTaster is web-based and free to use. The program has already become a common tool for the evaluation of deep sequencing projects, with more than 750 millions of variants analysed via our automatic pipeline for genotype files and up to 2 million single queries per day via the web interface. In addition to the evaluation of synonymous and non-coding alterations, MutationTaster is also more accurate than PolyPhen-2 and SIFT. The variants found in a complete exome can be analysed within 30 minutes, with a false positive rate of 1% using effective filtering strategies. With this project, we will extend our service to allow the accurate evaluation of all variants found in a Whole Genome Sequencing project within a few hours.

• DFG Programme: Research Grants
• Participating Institution: Cardiff University ; School of Medicine