Skip to main content
Premium Trial:

Request an Annual Quote

Positional Mapping Could Improve Success Rate of Exome Sequencing for Consanguineous Families


NEW YORK (GenomeWeb) – Clinical exome sequencing tests have diagnostic rates around 30 percent, but researchers at the King Faisal Specialist Hospital and Research Center in Saudi Arabia think those rates can be dramatically improved through a filtering strategy.

In a study published this month in Genetics in Medicine, the researchers described their strategy, which involves using positional mapping to narrow down the variant search, and found that they were able to identify likely causal variants in 88 percent of cases who had originally had a negative clinical exome.

Fowzan Alkuraya, a professor of human genetics at Alfaisal University and senior author of the study, told GenomeWeb that filtering strategies like positional mapping could help improve diagnostic rates of exome sequencing even more so than moving to whole-genome sequencing.

"Not knowing where to look in the exome makes it difficult to identify the right variant," he said.

Alkuraya said that the team was motivated to improve exome sequencing's sensitivity from their experience ordering clinical exomes from reference laboratories and getting diagnoses for only about 30 percent of cases.

In the study, he said the researchers first calculated the theoretical maximum yield of exome sequencing for individuals with a Mendelian phenotype that is mapped to a single locus, which they found could be as high as 95 percent.

To determine this hypothetical maximum sensitivity, Alkuraya said the team analyzed 104 families in depth. They were able to take advantage of the large family size and high rates of consanguinity to map the phenotypes of the affected individuals to a single locus. Then, he said, they looked in those regions for the causative mutation and in all but three cases, the mutation was found within the identified region. In addition, the variants identified were all genic, although 17 were splicing variants, including two that were more than 50 nucleotides away from the nearest exon. As such, the researchers determined that, in theory, exome sequencing should be able to identify 99 of the 104 cases, which would give it a maximum sensitivity of 95 percent.

Next, the researchers wanted to see whether they could use a positional mapping strategy to identify pathogenic variants in patients for whom they had previously ordered exome testing but had received negative results.

The group consented 33 families to a research protocol where they would resequence exomes and use positional mapping to narrow down the region where they looked for variants. The families were all consanguineous and had several affected members with the same autosomal-recessive phenotype.

The researchers first performed autozygome analysis for each family using a strategy they described in a previous study that involves first genotyping family members with a SNP array to identify autozygous intervals in patients. Autozygous intervals are regions of homozygosity from a common ancestor found in consanguineous families.

After sequencing affected members' exomes, the researchers used the auotzygome to filter variants, looking for causative variants only in the autozygous intervals shared between affected family members.

The researchers found causative mutations or likely causative mutations in 29 out of 33 families, or 88 percent, including mutations in both known and novel genes. When considering only variants found in known disease genes, they identified likely pathogenic mutations in 16 families, or 48 percent.

Looking at the variants that were missed by the clinical exome sequencing, the researchers noted that some variants did not appear to be "challenging" to call, such as missense variants in the ISCA2, ANSS, C3orf17, SLC1A4, and VRK1 genes.

In other cases they noted that the combination of positional mapping and exome sequencing was able to identify variants that would be hard to detect with exome sequencing alone.

One class of variants in particular that the team identified were splicing variants. Looking further at those variants, the authors noted that many of them were not the typical splice site variants and as such "are difficult to call with confidence as disease-causing by [exome sequencing] without supporting evidence from positional mapping."

However, David Adams, deputy director of clinical genomics at the National Institutes of Health's National Human Genome Research Institute and who was not involved in the study, said it was not clear from this paper whether those splice variants were actually pathogenic. Splice site variants' pathogenicity can be difficult to predict, he said, so unless the authors had previous knowledge about those specific splice sites, he said it is unclear whether they would be truly pathogenic.

Nonetheless, he said positional mapping was "a good approach to use" and a "powerful tool." Another aspect that should be taken into account, however, is that in consanguineous families affected by genetic disorders there is often more than one causative variant contributing to the disorder. 

Adams has also been evaluating ways to improve the diagnostic rates of exome sequencing as part of the NIH's Undiagnosed Disease Program. In a recent study, his team pointed to tools that would better enable the calling of medium-sized structural variants as potentially fruitful methods to identify undetected causative variants.

As for Alkuraya's positional mapping approach, Adams said that its utility may be limited to consanguineous families. Non-consanguineous families would have smaller regions of homozygosity, making it more difficult to identify autozygous intervals, he said.

Alkuraya agreed that autozygome mapping would not be as effective on non-consanguineous families, but said that the broader lessons about the location of pathogenic variants could be applied more generally. One takeaway, he said is that there needs to be more focus on improving filters so that causative variants are not thrown out. He noted the large number of splice variants that his group identified.

"It's difficult to predict splicing," he acknowledged, "but we need to invest in software that's better at predicting the effect of splicing." Another common filtering method is to get rid of variants that present in the population above a specific frequency, but variant frequencies can vary depending on the specific population, so a common variant in one group may be rare in another, he said.