SHANGHAI – Two-year-old Engine Biosciences Pte Ltd., a startup with an artificial intelligence (AI) platform approach to drug discovery, recently received a $10 million endorsement from Chinese and U.S. investors. Engine is one of a growing number of AI-first, data-driven biotechs that are plying the latest advances in machine learning to the complex task of understanding human biology to find new medicines.

Many are also increasingly making Asia their springboard. Engine is headquartered in Singapore, though it won't want for U.S.-based technical expertise. The startup has a team of four experienced academics, led notably by synthetic and network biology pioneer Jim Collins, the Termeer Professor of Bioengineering at MIT and Engine's scientific advisor.

The team also includes Hu Li, an assistant professor at the Mayo Clinic, trained in systems biology and computational biology. Li has developed machine learning software capable of uncovering novel regulatory mechanisms that explain the properties of biological phenotypes to benefit disease diagnosis, drug discovery and precision medicine.

Meanwhile Prashant Mali – Engine's other scientific co-founder and an assistant professor in bioengineering at the University of California San Diego – brings expertise developing combinatorial CRISPR screens for genetic interactions mapping. During his postdoc fellowship at Harvard's George Church Lab, he pioneered the development of the CRISPR/Cas systems for eukaryotic genome engineering.

But at the heart of the company are two American-Taiwanese brothers, Jeffrey and Timothy Lu.

Timothy Lu, a frequent collaborator of Collins, is a physician-scientist researcher with deep expertise in synthetic and systems biology, genetic interactions mapping, and computer science and engineering. He leads MIT's RLE Synthetic Biology Group and is also an entrepreneur, having co-founded Synlogic Inc., a company leveraging synthetic biology to genetically reengineer microbes. (See BioWorld Today, May 17, 2017.)

As CEO and board director, Jeffrey Lu brings his deep business acumen to Engine, having led data-driven companies in Asia for more than a decade starting as a consultant for Bain to driving AirAsia's (a Malaysian airline) data analytics transformation.

More than just AI

As Jeffery Lu explained to BioWorld, Engine's point of differentiation is not just the team's computational machine learning capabilities but their solution to the key issue facing any AI initiative: quality data. Without significant amounts of clean data, no AI project can succeed – the oft-described "garbage in, garbage out" conundrum.

For more than a decade, Collins along with his former student and also frequent collaborator, Hu, had been applying computational machine learning and deep learning toward modeling biological networks for drug discovery purposes – i.e. ways to identify new targets and ways to stratify patient populations as well as efforts in cellular reprogramming.

Along the way, they had to contend with the fact that there are few quality datasets available in the public domain for how biology is networked and wired. That is where Mali and Timothy Lu come in: Their wet lab biology tools leveraging CRISPR to do genetic interaction mapping helps Engine to control the quality of its data and outputs.

"We thought it would be very interesting to integrate what Jim and Hu have been working on; the ability to generate proprietary data to train algorithms to make predictions, then validate those in the wet lab and create a systematic feedback loop necessary to train the best predictive models," explained Lu.

"We are not a pure AI company," he added. "We feel strongly it is critical to have an internal platform or tool to generate our own proprietary data through biological experimentation that then trains our modeling to control the quality of that data."

Massively parallel

Given the complexity of cancer or neurodegenerative diseases, which are often caused by not just one but multiple genetic mutations, AI holds the promise of crunching through staggering amounts of data.

Engine's team is mapping biologic networks by computorially knocking out, knocking down or overexpressing genes in what it said is a massively parallel fashion. "The breakthrough is – instead of doing this kind of study in a serial one-by-one basis – we are now doing that in parallel and can do hundreds of thousands or even millions of tests at the same time in cellular models," said Lu.

The process includes building a screening library in the wet lab, which gets inserted into a population of cells to then track the effects of systematically knocking down various combinations.

"What happens when you knock out gene A and gene B and gene G. At the simplest level, do the cancer cells grow, die or does something else happen? The output of every experiment we do amasses hundreds of thousands, if not millions, of data points," said Lu.

The goal is to infer how the disease cell is wired genetically. But that early stage discovery process has several immediate applications.

First, Engine is using that drug discovery platform to develop its own pipeline of candidates in oncology, including liver cancer and neurodegenerative diseases, with early stage assets for Parkinson's disease.

The team also is open to collaborating with partners in areas outside of the company's core competence. Engine has a partnership with a Fortune 500 company in skin while for autoimmune and immunology, it is seeking partnerships.

Beyond doing early stage target discovery, its approach can also add value to drug repurposing and repositioning and to help with patient stratification for clinical trials.

Marriage of opposites

Engine is very much about marrying disparate things to create something new. Much like the marriage of engineering and biology that is behind synthetic biology – or AI with a wet lab – Engine has also taken the approach of bringing together the best that Asia and the U.S. can offer, with headquarters in Singapore and an office in San Francisco.

Engine set up in Singapore to benefit from a favorable corporate structure but has not strayed far from its academic roots. Engine is located in the Singapore-MIT Alliance for Research and Technology (SMART), a research enterprise established by MIT – where Collins and Lu call their academic home – and in partnership with the National Research Foundation of Singapore (NRF).

Looking at Engine's list of series A investors, the company also was able to tap into China's deep pool of investors.

The round was co-led by Silicon Valley-based DHVC (Danhua Capital) and 6 Dimensions Capital (formed through merger of Frontline Bioventures and Wuxi Healthcare Ventures in May 2017), with participation from global biopharmaceutical leader Wuxi Apptec, EDBI (a Singapore-based global fund), Pavilion Capital (a subsidiary of Temasek Holdings), Baidu Ventures, WI Harper and Nest.Bio Ventures.

China is fast emerging as a global AI powerhouse – thanks to government support, a deep pool of computer programmers, huge amounts of data and an open attitude to data privacy. Engine will likely find itself looking more toward China in the future.

"We are very interested in China. We have been working in liver cancer since the early stages of the company, in terms of extending that drug development program, and accessing more datasets in liver cancer. China is very important to us," said Lu. "On the operational side, now that we have closed up the funding and, given the strong China investors we have, we are exploring whether we operationally root further into China."

In the meantime, the company is looking to grow, and will double its headcount in the coming year. It is looking for people that can marry two disparate skills sets – those who understand AI with an understanding of biology.

Lu said it is not easy to find the right talent that understands both sides because in part "computer scientists tend to engineer things in a rational way, and to understand the rational underpinnings of systems whereas biologists recognize that biology has evolved over millions of years and is not necessarily structured the same way as a computer." Finding people skilled and interested in both is rare.

But when systems biology and the computer sciences come together, they can tackle one of the most complex questions: "How do we deconvolute how biology is wired and use that to enable a new way of thinking of drug discovery?"