Senjuti Saha was frustrated. As a microbiologist at the largest pediatric hospital in Dhaka, Bangladesh, she sees a lot of young kids with meningitis—inflammation of the membranes around the brain and spine. In up to 80 percent of those cases, she and her colleagues can’t work out which particular microbe is behind the condition. They draw samples of spinal fluids. They run simple tests. And usually: nothing.
Saha reasoned that she could find some answers through metagenomic sequencing—that is, analyzing the genetic material of everything in her samples and matching those readouts to databases of known microbes. The technique is becoming more common in the developed world, but it’s both technically and financially demanding. Saha spent more than a year trying to find the right collaborators, to no avail. Then she learned about Joe DeRisi.
DeRisi is a biochemist at the University of California, San Francisco, who has spent almost two decades developing ways of using genetic technology to fight infectious diseases. His team developed the ViroChip, a device for rapidly identifying viruses in bodily fluids, which they used to help identify the SARS virus in 2003. At San Francisco, he has worked to make metagenomics sequencing a regular part of hospital procedure—and that has already helped save people suffering from mysterious cases of meningitis and other brain infections.
But such work is expensive. Even if you can afford a sequencer and have the know-how to use it, you still need to compare the resulting reams of data against existing databases. And perversely, as sequencing gets cheaper and more people do it, those databases will swell, and “computing costs go up,” DeRisi says. “Our colleagues in Bangladesh aren’t going to spin up a big server farm.”
To take the brunt of this computational heavy lifting, DeRisi and his colleagues created an online platform called IDseq. It will crunch through any sequencing data that researchers upload, and tell them what’s in their patients. It’s free and based in the cloud, so it should be usable to anyone with an internet connection. The goal, DeRisi says, is “to enable people in under-resourced areas to do what we’ve been trying to do in San Francisco.”
Saha flew to California last March to learn how to do sequencing and to give IDseq a try. She brought with her 25 samples of spinal fluid from mystery meningitis cases. And after a couple of days, she discovered that three of those samples contained a virus called chikungunya. That was surprising, but it also made sense. The previous summer, Dhaka had experienced a bad chikungunya outbreak, and Saha remembers asking her colleagues if the infection could be partly responsible for the spate of meningitis cases. “[They] told me that chikungunya doesn’t cause meningitis,” she says.
The sequencing analysis suggested otherwise. Based on those results, Saha developed a low-cost test for chikungunya and analyzed 478 more meningitis samples back in Bangladesh. Of those, 17 tested positive for the virus, and they had all been collected precisely when the chikungunya outbreak was raging through Dhaka. Perhaps this virus causes neurological problems more commonly than anyone thought. “It’s a poster child for what we want to do with IDseq,” says DeRisi.
“Senjuti’s success got us all excited,” says Cori Bargmann, the president of science at the Chan Zuckerberg Initiative—a new company founded by the pediatrician Priscilla Chan and her husband, Facebook’s Mark Zuckerberg, that has pledged $3 billion toward fighting infectious diseases, and is footing the bill for IDseq’s software development and computational power. “It was a big win, and it was critical that she was the one doing it with her own samples. Global-health workers will have to be able to do this themselves.”
To be clear, this isn’t a new idea. There are already other tools that perform similar jobs, and while some require subscriptions, several are free. One system, known as ViralNGS, was used extensively during past outbreaks of Ebola and Lassa fever. Another, known as Nextstrain, allows scientists to compare viral sequences and track evolving outbreaks in real time.
DeRisi argues that IDseq is different in that it was crafted not by academics, but by a large team of software engineers, security experts, and more. “It is industrial-strength commercial software,” he says, and should be easier to both scale up and keep running. The team has also worked to fit IDseq with an intuitive and user-friendly dashboard so it can be used by people without advanced technical expertise. “I wasn’t able to find anything else that’s free and has a good interface,” Saha says.
So far, DeRisi’s team has also used IDseq to analyze bodily fluids from feverish Ugandan children. Amid the several predictable cases of malaria, they found a rogue’s gallery of respiratory viruses, including one that was new to science. They have started working with health officials in California’s Alameda County to identify viruses in local mosquitoes. And they’re now ready to roll out the platform more widely.
On Tuesday, at a meeting in Berlin, the Bill and Melinda Gates Foundation announced a new training scheme for clinicians around the world. The foundation will bring successful applicants to the Chan Zuckerberg BioHub labs in San Francisco, where DeRisi is co-president, so they can learn how to use IDseq from his team. Each clinician will get a backpack-size sequencer, a year’s worth of chemical reagents, and technical support so they can start doing sequencing in their home countries.
DeRisi’s grand plan is to turn IDseq into a common dashboard, shared by disease detectives around the globe. If enough people use it, they could theoretically start spotting the movements of diseases across borders, the emergence of new illnesses, or the spread of drug-resistant strains. “That will be incredibly valuable,” says Jennifer Gardy, an epidemiologist at the University of British Columbia who is not connected to the project. “Information sharing is one of the most powerful public-health interventions in an outbreak.”
But Gardy adds that there’s still a lot of work to be done. “It’s easy for us to sit in our labs dreaming up tools and platforms, but we need to make sure we’re designing them in a way that makes sense to the doctors, nurses, lab techs, and epidemiologists out there in an outbreak,” she says.
Paula Fernandes of GSS Health, who is currently assessing different diagnostic technologies, is skeptical about how ready IDseq is for use in the field. Even if it’s easier to analyze the results of metagenomic sequencing, the technique itself still demands bulky equipment and a lot of expertise. “It’s not at all remote-friendly,” she says. And there’s a history of scientists’ getting overly excited about “promising technologies, which then simply sit on shelves in poorly resourced labs gathering dust long after glossy global-health reports have been filed.”
There are other limitations, too. Saha says it can be hard to acquire the chemical reagents used in sequencing experiments, because of issues with customs and unreliable supply chains. Once those experiments are complete, health workers would also have to upload their data to IDseq—a challenging feat in rural areas with slow connection speeds. “If the transfer is too slow or the data too large, we just FedEx hard drives,” she says.
DeRisi says none of these problems are deal breakers. Sequencers are getting smaller and cheaper. From his experience, anyone with basic laboratory skills can be trained to use them. His team is working on ways of compressing sequencing data so it can be more easily uploaded.
There’s also the matter of interpretation. Let’s say you use IDseq and it churns out a list of microbes. What next? Some viruses are glaringly obvious causes of disease, but Fernandes notes that many microbes are more ambiguous. Staphylococcus can cause severe infections, but many people have it up their noses. Someone with a fever might have the plasmodium parasites that cause malaria, but the latter doesn’t necessarily explain the former. Metagenomic sequencing tells you who’s there; it doesn’t tell you what they’re capable of.
DeRisi acknowledges this, but he hopes that as more people use the platform, it will become easier to connect the dots between microbes and symptoms, and “make meaningful judgments about the worth of any particular finding.”
“IDseq is an excellent tool, but it needs to be paired with people who have substantive knowledge to guide its use,” says Saha. “It doesn’t solve all the problems, but it means that groups like ours don’t have to spend time to build up [sequencing] capacity. And anything is better than nothing.”