SALT LAKE CITY (GenomeWeb) – Researchers from the Children's Hospital of Philadelphia are exploring the feasibility of using crowd sourcing to help track down and interpret secondary, or incidental, findings in whole-exome sequences for children assessed through CHOP's pediatric sequencing program, PediSeq.
CHOP researcher Ian Slack presented preliminary findings from this foray into variant interpretation by so-called "PediSeeker" volunteers at the annual American College of Medical Genetics and Genomics meeting here last week.
The PediSeq project itself is intended to determine best practices for exome sequencing across several different realms in the pediatric population, Slack explained.
Along with trying to decipher disease-related variation in the pediatric protein-coding sequences, the PediSeq group is also providing secondary findings to pediatric patients' clinicians in cases where parental consent has been obtained prior to exome sequencing.
The search for the secondary findings encompasses the list of 56 genes recommended for incidental findings return by the ACMG, Slack noted, along with more than 2,500 genes that CHOP believes may be medically actionable.
Although bioinformatics pipelines have made it possible to routinely whittle down the 900,000 or so variants identified in each sequenced exome, on average, the final 20 candidate variants for secondary findings still tend to create an analytical bottleneck because they must be checked manually.
In the interest of widening that bottleneck, members of the PediSeq team decided to look beyond the core group of experts already involved in the pediatric project by tapping into collective, volunteer efforts from other parts of CHOP.
The researchers recruited 10 such PediSeeker volunteers from CHOP — a group that included individuals who are knowledgeable about genetics, as well as at least one person with prior variant interpretation experience.
The PediSeekers' variant interpretation assignments were restricted to secondary findings, Slack noted, because these variants are not only less prone to inadvertently identify the case involved, but also tend to be associated with a higher level of published evidence supporting their role as being pathogenic, non-pathogenic, or somewhere in between.
After assigning each volunteer more than a dozen variants to analyze, the PediSeq members got feedback and findings back from the volunteer team. They then used this information to evaluate everything from the workload associated with analyzing these variants to the accuracy of the PediSeekers' secondary finding interpretations, as compared with calls by PediSeq experts.
Following the first 170 variant interpretations by the PediSeekers, Slack and his colleagues did a secondary review of the interpretations. Forty-five of the interpretations have undergone secondary review so far, while 125 are yet to be reviewed.
Some two-thirds of the first 45 variant interpretations considered received identical calls by both the PediSeekers and the primary PediSeq analysis team. When differences did arise, they tended to be first degree discordance, meaning the variant interpretations were off by just one level (i.e., "likely benign" rather than "benign," and so on).
Moreover, there were only two instances of discordant interpretations that would actually impact the information that's returned to clinicians about their pediatric patients.
That reflects the level of confidence in a pathogenic call that's required for returning incidental findings, Slack explained. For the two cases where PediSeekers made calls that altered secondary results return, the volunteers were more apt to interpret variants as being pathogenic than were their expert PediSeq peers.
Several sources of interpretation discordance remain in general, including a dearth of concrete evidence for some variants and differences in allele frequencies in various populations.
Nevertheless, Slack argued that preliminary results suggest that variant interpretation crowd sourcing "offers certain efficiencies," while providing hands-on experiences to individuals at CHOP who may not be involved in genomic research at the moment, but have an interest in trying their hand at genomic analysis.