This is approved reprint article of GenomeWeb’s Saphetor’s VarSome Search is Latest Step on Road to Community of Genomics Professionals, published on June 26, 2018, by Neil Versel.
Saphetor’s VarSome Search is Latest Step on Road to Community of Genomics Professionals
CHICAGO (GenomeWeb) — In May, Swiss precision medicine analytics company Saphetor SA released an update to its VarSome genomic annotation tool, featuring, for the first time, a full search function.
“You can search for any term that appears anywhere on VarSome, including any publications linked to it,” said Saphetor CEO Andreas Massouras. “It’s a bit of the Google of genetics if you like. It connects everything to everything else.”
However, Massouras was careful to say that VarSome’s search is more targeted than Google’s. “You can search for the same thing on Google,” but get many extraneous or irrelevant results, he explained.
Technology called MolecularDB is behind both VarSome and Saphetor’s variant analysis platform for next-generation sequencing. “It’s a technology that allows very fast indexing for genomics data, and fast retrieval,” Massouras said.
“On the back of that technology, we’ve built our variant analysis platform,” as well as an application programming interface. The VarSome website is one special usage of the API. VarSome, which pulls from public databases as well as user-contributed information, displays data in an aggregated manner and attributes results to their original sources. “We link everything in our database to everything else,” Massouras said.
“If you say ‘cystic fibrosis,’ you will get 11 genes, 58 diseases, 109 phenotypes, 1,600 articles, and so on,” he explained. “You can click on any of those to narrow down your search.” Doing so brings up not only the source article, but a list of variants connected to each article. Clicking on a variant takes the user to a variant landing page. “Most of the queries on VarSome are for specific genetic variants,” Massouras noted.
“It connects the dots and shows very comprehensive data,” he continued, or at least as comprehensive as medical knowledge allows. “Many of these variants have not themselves been classified because they are unique or rare,” Massouras said.
Users also can add comments and contribute educational content to VarSome. “We pull in the abstract and the authors of this publication,” Massouras said while demonstrating a user-contributed listing. “This is what the user said about this variant.”
Lausanne, Switzerland-based Saphetor continues to develop this connectivity and similar “community-based features,” according to Massouras. He said that users have been asking for ways to connect like-minded people, such as those looking for variants related to rare diseases. “The more features and data we add to it, the more it becomes the go-to place for genomics queries,” Massouras said.
Two months before the introduction of the search function, Saphetor launched an automated pathogenicity classifier in VarSome, based on American College of Medical Genetics and Genomics (ACMG) classification guidelines.
In February, Saphetor partnered with Genomenon, an Ann Arbor, Michigan-based maker of a genomic literature search engine, to share genomic variant information between their data platforms.
This came just weeks after the company released an API for VarSome. Saphetor said that the API can increase annotation throughput by fourfold compared to other methods, helping researchers process a full genome in about two hours over an “average” internet connection.
The next feature planned is the ability to see future viewers of a page. “If you yourself are a patient with a rare mutation or if you have a patient with a rare mutation, you can declare interest and try to connect with other patients around the world with the same mutation,” Massouras said. The “critical mass” of users and community-based approach, he believes, gives Saphetor a chance of succeeding in an area where others have failed.
“This can be an extremely valuable service because it will help research and, in the end, the clinical people,” Massouras said.
Saphetor also plans on introducing more connectivity between VarSome community members and to pharmaceutical companies. “Based on the query, we [will be able to] identify potential fits between a patient and a drug being developed,” Massouras said.
Other advanced features in the pipeline include expansion of the databases it aggregates and automated classification to include more data and make it more accurate. “Our job is never done. There are always improvements to be made,” Massouras said.
VarSome has been growing at a rapid pace. Since it came online in May 2016, the user base has grown to more than 62,000 worldwide in 163 countries, though three-quarters come from either the US or Europe. That total has been expanding at nearly 24 percent per consecutive month since inception, and the number of users has approximately quadrupled year-over-year in each month of 2018 so far, Massouras reported.
Nearly 60 percent of users are clinical laboratory professionals or physicians working in diagnostics, 30 percent are academic researchers, and most of the rest are from commercial labs and genomics companies. Massouras said that about 70 percent found VarSome through personal referrals, while 16 percent came from search engines and 14 percent from online referrals.
“We are reaching out more to clinicians who would like to learn about genetics,” Massouras said. “VarSome is used routinely in medical school lectures. It’s also used in seminars for trained physicians who want to learn about genetics.”
Bioinformaticians mostly use the Saphetor API rather than the VarSome website, since the API can accept queries of up to 50,000 variants each. Massouras reported that API usage is in fact about six times higher than VarSome’s.