Skip to main content
Premium Trial:

Request an Annual Quote

Precision Medicine Informatics Startup Jungla Emerges From Stealth Mode


CHICAGO (GenomeWeb) – A newly emerged bioinformatics and biotechnology company has launched a technology platform that with the help of artificial intelligence continuously integrates clinical knowledge with genomic and biophysical data to create molecular evidence for both research and clinical purposes.

The company, called Jungla, emerged from stealth mode on Nov. 27 at a Hewlett Packard Enterprises event in Madrid by unveiling what it calls the Molecular Evidence Platform.

"We're not just a computational platform," explained CEO and Cofounder Carlos Araya. "We're a computational and an experimental platform. We are mechanistic. The computational models are not just AI. There's more than one mechanism going into them."

In pre-launch testing, two-year-old Jungla was able to show, proactively, positive predictive values and negative predictive values of more than 95 percent for 4 million human mutations and variants deemed clinically relevant, according to Araya. Various open-source tools that Jungla tested against its own technology showed PPVs of about 75 percent and NPVs of approximately 85 percent, he said.

"We are characterizing more and more [how] the molecules and the cells out there that are affected by mutations," Araya said. This means that Jungla is able to say with high confidence whether or not a mutation is pathogenic, he explained.

Headquartered at Johnson & Johnson Innovation's JLabs @SSF biotech incubator in South San Francisco, California, Jungla is funded by $2.5 million in seed funding from Andreessen Horowitz that dates to late 2016.

In addition to Araya, Jungla was cofounded by Jason Reuter, a former Stanford University postdoctoral researcher in functional genomics, and Alexandre Colavin, who earned a doctorate in biophysics at Stanford. All three left the school in by 2017 to start the company, but the idea goes back a decade, when Araya was a working on a PhD in genome sciences at the University of Washington; he later did postdoctoral work at Stanford.

Araya did his PhD and some postdoc research on large-scale functional assays. Starting in 2008, he led development of a technique called deep mutational scanning, a technique that couples mutagenesis of proteins and their selection with high-throughput analysis of mutations by next-gen sequencing.

"It allows you to measure the effects of hundreds of thousands of versions of the protein in a single assay in parallel," Araya explained. "Deep mutational scanning has become kind of a staple of protein science now."

Araya made his way to Stanford in 2011, where he became a postdoctoral researcher in genomics and biophysics. "I felt strongly that genomicists were not well equipped to develop the solutions needed to understand the effects of mutations," he said.

So Araya and several colleagues in Palo Alto, California, advanced his earlier work by actually breaking open Illumina sequencers to take over the fluidics, optics, temperatures, and biochemistry and modify the machines to process large-scale physical assays.

"We could do 10 million biophysical assays in one single experiment," he said.

Next, he started to look at software to leverage this "genomic-scale information," as Araya described it. Around the same time, a relative was diagnosed with stage IV melanoma, so Araya developed an interest in creating new analysis techniques for cancer genomes. "That gave me the opportunity to interact with the clinical space," he said.

"Commercial labs, they were scaling up tests by a lot, but they really were not built to either develop or even really effectively use technologies like the ones that we have been developing to inform on the effects of mutations at scale," according to Araya. "They really weren't going to be able to make the most of the data that was being acquired."

Araya, Reuter, and Colavin decided that they could make better technology — and do so at scale — to help scientists better understand the effects of mutations. Thus, Jungla, named for the Spanish word for "jungle," has set out to help researchers and clinicians alike sort through and make sense of this mass of data that genomics has created.

The name, according to Araya, comes from the founders' belief that genomes really are not like a sequence, despite the common process called sequencing. "We actually feel like they're like an ecosystem where you have many different molecules working together," Araya said. "We need a diversity of different areas of expertise and domains to tackle these problems."

Araya said that other genomic technology companies are missing an opportunity by viewing sequences simply as strings of letters, then trying to figure out how mutations change these strings, and then correlate the changes to progression of symptoms and patient outcomes.

"We don't pursue that strategy," Araya said. "We're much more of the view that to really understand the effects and be able to make meaningful predictions, we need to be able to not just think of these mutations as something that changes the sequence, but really think of them like how do they change the behavior of the molecules and the cells where they occur."

Jungla prefers to focus on molecular-level and cellular-level models from both the computational and experimental perspectives to, in the words of Araya, "inform on the effects of mutations at scale, learn from those patterns, and then make predictions."

To this end, Jungla produces and intends to distribute models to hospitals and other clinical organizations.

"We are trying to arm [clinical teams] with the best solutions for interpreting mutations from a technical point of view, doing this quantitatively so we can move the field from observational to quantitative models of interpretation," Araya said. "We see ourselves as kind of a partner in the larger ecosystem where there are many different entities with different areas of expertise contributing to the health of the patient."

Jungla is now in its third phase of development. The company spent its first six months of stealth operation developing its business strategy, building proof-of-concept tests, and raising money. For most of the last two years, Jungla focused on further developing its technology, from both the computational and lab sides.

"Over the last half year, we've also started working with some target clients," Araya said, adding that the firm expects some customer announcements "soon."

The first such announcement was with HP at the recent Madrid event. Jungla is turning to large-scale HP hardware acceleration technology called "memory-driven computing" to improve processing speed.

"We're going to invest heavily into the tools that play important roles in the precision medicine foundation," Araya said.

He said that it could take 250 days to compare massive files of genetic mutations to known genetic correlations with traditional analysis software on standard computer platforms. With Jungla computing techniques and the HP infrastructure together providing 250X acceleration, Araya said that this could be cut to a single day.

"That brings it to a timeline that is clinically feasible," he said.

Jungla has not yet loaded live patient data onto the HP platform, but Araya said that the company should be able to demonstrate clinical outcomes by mid-2019.