By Deborah Borfitz
June 26, 2019 | One of the most vexing problems in diagnostics, crossing multiple specialty areas, is the large number of rare diseases that can take years, if ever, for a physician to identify. In pediatrics alone, genetic variants have been tied to more than 2,000 rare genetic disorders that most physicians have never heard, says David FitzPatrick, professor at the Medical Research Council Human Genetics Unit at the University of Edinburgh and a practicing pediatrician. Yet one in 300 babies has one of those disorders, making them common on a population level.
Answers may lie in human DNA, but until recently there was no efficient way to sort through the four to five million genetic variations to find the one or two causing problems, FitzPatrick says. New point-of-care DNA analysis software is now freely available that could change all of that, making genetic conditions easier to diagnose in clinical practice and research programs.
FitzPatrick was part of the research team that developed the new tool, called VEP-G2P (Variant Effect Predictor-Gene2Phenotype), described in a recently published article in Nature Communications (https://doi.org/10.1038/s41467-019-10016-3). The University of Edinburgh collaborated with the European Bioinformatics Institute (EBI) on the project, with funding from biomedical research charity Wellcome Trust. VEP-G2P quickly identifies likely disease-causing genes based on knowledge encoded in a database hosted by EBI, he says.
The EBI provides access to the software, which is designed for use by diagnostic laboratories and physicians who are "reasonably competent" at dealing with genetic data, FitzPatrick says.
The closest comparable tools currently available are Stanford's AMELIE (Automatic Mendelian Literature Evaluation) and Congenica's Sapientia, he says.
Filtering Power
While the volume of genetic data that gets filtered by the software is enormous, the process is "incredibly simple," FitzPatrick says. Common variants that are typically not related to disease, as captured by databases such as the Broad Institute's gnomAD, get eliminated. So do rare variants in parts of the genome that don’t cause problems. That leaves rare variants that code for proteins, which get coupled with disease-specific information such as known genetic associations and whether one or both copies of the gene change are needed to cause disease, he explains.
In many cases, the search is for a single, disease-causing disruption in the genome out of the six billion base pairs of DNA per cell, FitzPatrick says. It is rare for more than four variants to come through the filter.
The tool is intended to "democratize" whole-genome and whole-exome diagnostic approaches, particularly for healthcare systems in economically-developing countries that have access to some sequencing data, he says. It could help clinicians quickly arrive at a reasonably-accurate diagnosis of complex developmental disorders and genetic eye and skin diseases, as well as to characterize pathways activated by somatic cancer mutations and to better describe cancer predisposition—"any specialty where there is a very strong number of genetic disorders, which is most clinical specialties."
Even with this tool, only about half of all pediatric cases can be diagnosed because "we don't know what the other half of disorders are caused by," says FitzPatrick. "But that's much better than five years ago when we could diagnose maybe 1 in 20 of them [by any means]."
The software can also predict consequences of DNA changes even without the link to a known condition, he continues. In pediatrics, most genetic variants are new, first-time occurrences rather than inherited. "You don't need to have seen the variant before to make a diagnosis as long as it’s in the right gene and has the right consequence," based on what is known about the disease.
On the most powerful capabilities of VEP-G2P is identifying false-positive variants that can confuse diagnosticians because they "look plausible but are just not the cause [of the problem]," FitzPatrick says.
Faster Answers
The tool is currently being used by FitzPatrick in both a clinical practice and research capacity. The Sanger Institute in Cambridge is also an early adopter.
The value for researchers is in more easily finding and qualifying patients for genetic disease trials, he notes. The don’t have to go through, for example, 300 variants to identify the few that are relevant for a specific patient.
Using VEP-G2P, FitzPatrick was recently able to diagnose a boy with Christianson syndrome (caused by mutations in SLC9A6) after 14 years of trying unsuccessfully on his own. "I've known him since he was a baby and he always had very severe developmental problems and was fairly dependent on his parents." The answer came nearly as fast as his genome could be sequenced, an enormous relief to his parents who had been putting off having other children out of fear the same thing might happen again. They now know that is very unlikely.
Deciphering Developmental Disorders (DDD), a largescale study of children from the UK and Ireland with severe developmental disorders, provided the dataset to validate the tool—and it passed with flying colors, reports FitzPatrick. "The tool could identify all of the variants that we thought it should be able to identify."